Natural History of the Hyperinsulinism/Hyperammonemia Syndrome: A Retrospective Review Incorporating Patient-Centered Data

Abstract

Introduction

The hyperinsulinism/hyperammonemia (HI/HA) syndrome manifests with fasting and protein-induced hypoglycemia, hyperammonemia, and neurodevelopmental features including epilepsy. There is a paucity of information describing the natural history of the HI/HA syndrome.

Methods

A retrospective review of patients with HI/HA syndrome evaluated at the Congenital Hyperinsulinism Centers at the Children’s Hospital of Philadelphia or Cook Children’s Medical Center or who contributed to the Congenital Hyperinsulinism International HI Global Registry (HIGR) was conducted to describe the natural history of the HI/HA syndrome with particular focus on treatment and neurodevelopmental outcomes.

Results

A total of 66 patients (36 female) from the medical record review and 15 patients (7 female, 3 sex not reported) from HIGR were included. Median age at last follow-up was 13.1 years (IQR, 6.8–19.0 years) in the medical record cohort and median age at survey completion was 11.6 years (IQR, 5.5–18.0 years) among HIGR participants. Eighty percent of the medical record cohort and 82% of HIGR participants were treated with diazoxide and most continued treatment over the follow-up period. Epilepsy (29% of medical record cohort, 33% HIGR participants) and neurodevelopmental issues (64% medical record cohort, 73% HIGR participants) were common.

Conclusion

Our findings underscore the importance of longitudinal endocrine and neuropsychological follow-up for patients with HI/HA syndrome and demonstrate the potential value of patient-driven registry data to address persistent knowledge gaps in the natural history of rare diseases.

Introduction

Hyperinsulinism/hyperammonemia (HI/HA) syndrome (OMIM 606762) is the second most common cause of congenital hyperinsulinism [1]. HI/HA syndrome is clinically characterized by fasting and protein-induced hyperinsulinemic hypoglycemia, hyperammonemia, and distinct neurologic manifestations. It is caused by activating heterozygous pathogenic variants in GLUD1, encoding glutamate dehydrogenase (GDH) [2]. To date, disease-causing variants in the GLUD1 gene have been identified in exons 6 and 7, encoding the GTP allosteric binding domain, and in exons 10, 11, and 12, encoding the alpha-helix and antenna-like domains [3, 4]. GDH is highly expressed in pancreas, liver, kidney, and brain and catalyzes the reversible conversion of glutamate to alpha-ketoglutarate and ammonia [5]. In the pancreatic beta-cell, alpha-ketoglutarate is oxidized via the tricarboxylic acid cycle leading to generation of adenosine triphosphate (ATP), closure of ATP-sensitive potassium (K_ATP) channels and insulin secretion. Increased flux through GDH in HI/HA syndrome leads to increased basal insulin secretion as well as increased protein-stimulated insulin secretion because the amino acid leucine is a GDH activator [6]. Hypoglycemia in HI/HA syndrome is typically well-controlled with the K_ATP channel agonist diazoxide.

Ammonia levels are usually elevated 2–5 times the upper limit of normal, although individuals with normal ammonia levels have been described [7, 8]. Hyperammonemia results from GDH overactivity in the kidney, does not respond to ammonia lowering therapies, and does not appear to cause symptoms [9]. Epilepsy occurs more commonly in HI/HA syndrome than in other forms of hyperinsulinism and is characterized by a unique seizure phenotype. These atypical absence seizures occur in the setting of euglycemia and are distinct from the focal-onset seizures that may occur following hypoglycemic brain injury [10, 11]. Developmental delays, learning disorders, and attention deficit/hyperactivity disorder (ADHD) have also been described to be more prevalent in HI/HA syndrome than in other forms of hyperinsulinism [7, 11, 12]. These neurologic manifestations are incompletely explained by prior hypoglycemia and are hypothesized to result from aberrant GDH activity in the brain [12].

Cochrane and colleagues likely provided the first clinical description of HI/HA syndrome in their report of 4 patients with profound leucine-sensitive hypoglycemia in 1956 [13]. In 1998, the molecular etiology of the HI/HA syndrome – activating pathogenic variants in GLUD1 – was elucidated by Stanley et al. [2]. Since then, a number of case reports and case series describing the clinical, biochemical, and genetic characteristics of HI/HA syndrome have been published. While the phenotype of this disorder has been well described, there remains a paucity of data describing its natural history. Consequently, disease trajectories among patients with HI/HA syndrome remain incompletely understood. In this study, we aimed to describe the natural history of the HI/HA syndrome combining medical record review from two large congenital hyperinsulinism referral centers with patient- and caregiver-reported data from the Congenital Hyperinsulinism International (CHI) HI Global Registry (HIGR), an international, patient-powered hyperinsulinism registry.

Methods

This was a retrospective study combining data extracted from the medical records of patients evaluated at the Children’s Hospital of Philadelphia (CHOP) Congenital Hyperinsulinism Center and the Hyperinsulinism Center at Cook Children’s Medical Center (CCMC) with HIGR data to describe the natural history of the HI/HA syndrome with particular emphasis on hyperinsulinism treatment and neurodevelopmental course. The protocol for this study was approved by the Children’s Hospital of Philadelphia Institutional Review Board (IRB), Protocol No. 22-019652; Cook Children’s Medical Center IRB, Protocol No. 2022-005; and North Star Review Board IRB (which reviewed and approved the HIGR research study), Protocol No. NB100012, respectively. HIGR is an international registry for patients with hyperinsulinism and their caregivers to self-report information on the diagnosis and management of hyperinsulinism, comorbidities, and quality of life via online surveys [14]. All HIGR study participants provided informed consent to participate in research when they joined the registry. The Children’s Hospital of Philadelphia IRB and Cook Children’s Medical Center IRB granted exemption from requiring written informed consent for medical record data abstraction for this study, respectively.

Inclusion criterion for all data sources was a diagnosis of HI/HA syndrome. For the medical record cohort (CHOP and CCMC data sources), the diagnosis of HI/HA syndrome was based on genetic testing findings of a pathogenic variant in GLUD1 or biochemical evidence of hyperammonemia and hyperinsulinism, as described previously [15]. In HIGR, the diagnosis of HI/HA syndrome was based on self-report of a pathogenic variant in GLUD1. Patients were included if they were evaluated at CHOP between January 1998 and November 2022, at CCMC between October 2010 and November 2022 or contributed survey data to HIGR between October 2018 and October 2022. Start dates for patient inclusion in the cohort were based upon the start dates of the congenital hyperinsulinism programs at CHOP and CCMC, respectively, and the start date of data collection in HIGR.

The following demographic and clinical data were extracted from all sources: age, sex, race, ethnicity, gestational age, birth weight, age at first presenting symptom, symptoms at presentation, hyperinsulinism treatment, seizure history, and neurodevelopmental history. Age at diagnosis of hyperinsulinism, age at last contact, ammonia level, oral protein tolerance test results, and genetic testing results were abstracted only from the medical record. Quality of life data were extracted only from HIGR since these data are not routinely collected as part of clinical practice and were absent from the medical record.

Race was categorized as American Indian or Alaskan Native, Asian, black or African American, Native Hawaiian or Pacific Islander, white, greater than one race, other, or unknown/not reported. Ethnicity was categorized as Hispanic or Latino, not Hispanic or Latino, or unknown/not reported. At the time of data extraction from HIGR, race and ethnicity data were collected only for participants in the USA and were therefore coded as “unknown/not reported” for HIGR participants residing outside of the USA. Gestational age and age at first presenting symptom were extracted as continuous variables from the medical record and categorized to align with data collection in HIGR. Gestational age was categorized as: preterm (<37 weeks gestation), term (37–42 weeks gestation), and post-term (>42 weeks gestation). Age at first presenting symptom was categorized as: 0–6 days, 1–4 weeks, 5 weeks–6 months, 7–11 months, 1–3 years, and >3 years. Patients in the medical record cohort were operationally classified as having epilepsy if they had a documented epilepsy diagnosis or were prescribed antiseizure medication. HIGR participants were classified as having epilepsy if they responded “yes” to the survey item: “Has the participant been diagnosed with epilepsy (recurrent seizures not from blood sugar).” At CHOP and CCMC, adequate control of hypoglycemia is assessed based upon review of home blood glucose monitoring data and periodic inpatient evaluation, including provocative fasting tests and oral protein tolerance tests, typically every 1–2 years. Diazoxide doses are titrated to attain age-appropriate overnight fasting tolerance and to ameliorate protein-induced hypoglycemia. Protein sensitivity on oral protein tolerance testing was defined as nadir plasma glucose <2.8 mmol/L (<50 mg/dL) or >0.7 mmol/L (>12 mg/dL) decline in plasma glucose from baseline, as previously described [16].

Statistical Analysis

Data from the medical record cohort were reported separately from the HIGR cohort data to eliminate potential overlap between sources. Cohort characteristics were summarized using descriptive statistics, including counts and percentages for categorical variables, means and standard deviations (SDs) for normally distributed continuous variables, and median and interquartile range (IQR) for nonnormally distributed continuous variables. The number of HIGR responses was reported separately for each variable because HIGR participants may not complete all surveys or all questions within a given survey. To describe general population level trends in diazoxide dose requirement across the lifespan, we calculated the mean diazoxide dose for age by collapsing the dataset to a single observation per each year of age. Due to the limited number of repeated dose measurements available per individual in this retrospective dataset, longitudinal modeling of within-individual diazoxide dose trajectories was not performed. To compare age at last recorded diazoxide dose between categories of diazoxide treatment intensification and de-escalation over the follow-up period for patients in the medical record review cohort Kruskal-Wallis test was used and post-hoc pairwise comparisons were conducted using Dunn’s test with Bonferroni correction. To explore genotype-phenotype associations, clinical features were compared between patients with pathogenic variants in exons 6 and 7 of the GLUD1 gene (encoding the GTP allosteric binding site) and those with pathogenic variants in exons 10, 11, and 12 (encoding allosteric regulatory domains). These analyses were performed both within our medical record cohort and in a pooled dataset that included previously published patients with HI/HA syndrome [7, 8, 11]. Wilcoxon rank-sum tests were used to compare medians of nonnormally distributed continuous variables and χ² tests were used to compare proportions between groups. Two-sided p value <0.05 was set as the threshold for statistical significance. Statistical analyses were performed using Stata 18 (Stata Corp., College Station, TX, USA).

Results

A total of 66 patients (36 females, 30 males) with HI/HA syndrome were identified via medical record review (n = 61 CHOP, n = 5 CCMC). Fifteen patients (7 females, 5 males, 3 sex not reported) with HI/HA syndrome were identified via HIGR. The cohort was predominantly White and not Hispanic or Latino, noting that demographic data were unavailable for many in both cohorts (Table 1). Median age at last follow-up was 13.1 years (IQR, 6.8–19.0 years) in the medical record cohort and median age at the time of survey completion was 11.6 years (IQR, 5.5–18.0 years) among HIGR participants (n = 12).

Table 1.

Cohort demographics

​	Medical record (n = 66)	HIGR (n = 15)
Race, % (n)
Asian	1.5 (1)	0 (0)
Black or African American	4.5 (3)	0 (0)
White	62.1 (41)	33.3 (5)
Other	15.2 (10)	13.3 (2)
Unknown/not reported	16.7 (11)	53.3 (8)
Ethnicity, % (n)
Hispanic or Latino	12.1 (8)	6.7 (1)
Not Hispanic or Latino	56.1 (37)	13.3 (2)
Unknown/not reported	31.8 (21)	80.0 (12)

Among those for whom birth history was available, the majority were born at term (90%, n = 57 medical record cohort; 73%, n = 11 HIGR responses) with normal birth weight (median birth weight 3.2 kg [IQR, 2.8–3.8 kg], n = 54 medical record cohort; 3.6 kg [IQR, 2.9–4.3 kg], n = 11 HIGR responses).

Age and symptoms at presentation are summarized as shown in Figure 1. In both cohorts, approximately 30% of patients with HI/HA presented in the first week of life, with seizure as the most common presenting symptom. However, late recognition of hypoglycemia was also frequent, with onset of hypoglycemic symptoms after 6 months of age in 39% of the medical record cohort (n = 61) and 21% of HIGR participants (n = 14 responses) for whom data on presentation were available. Twenty percent of the medical record cohort and 14% of HIGR participants (n = 14 responses) were initially diagnosed with epilepsy prior to receiving a diagnosis of hyperinsulinism. In the medical record cohort, median age at presentation was 5.0 months (IQR, 2 days–12.0 months) and hyperinsulinism was diagnosed at a median age of 6.0 months (IQR, 1.3–15.1 months) with a median time from presentation to hyperinsulinism diagnosis of 30 days (IQR, 0–112 days).

Fig. 1.

Distribution of age (a) and symptoms (b) at presentation in the medical record cohort (blue) and HIGR (orange). Patients could have more than one presenting symptom. AMS, altered mental status; d, days; HIGR, HI Global Registry; m, months; w, weeks; y, years.

Molecular genetic data were available for 65 patients in the medical record cohort. Inheritance of the GLUD1 pathogenic variant was de novo in 46%, familial in 25%, non-maternal (paternal sample unavailable) in 8%, and unknown in 22%. Pathogenic GLUD1 variants were in exons 10, 11, or 12 in 57% of the cohort and in exons 6 or 7 for the remainder (online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000551535). Median ammonia level was 98.2 μmol/L (IQR, 79.1–142 μmol/L, n = 42, normal range: <33 μmol/L). Oral protein tolerance test results were available for 39 patients and demonstrated protein sensitivity in all but 2 patients, both of whom were treated with diazoxide at the time of testing. These data were not collected in HIGR.

Hyperinsulinism Treatment and Course

Diazoxide was the most common treatment and was used to treat 80% of the medical record cohort and 82% of HIGR participants (n = 11 responses). Seven patients in the medical record cohort (11%) were treated with near-total pancreatectomy, all of whom were born prior to identification of GLUD1 pathogenic variants as the cause of HI/HA syndrome. Median age at the time of pancreatectomy was 2.5 years (IQR, 0.5–5.9 years). None of the HIGR participants reported history of pancreatectomy. Four patients in the medical record cohort (6%) and two HIGR participants (18%, n = 11 responses) were treated with dietary intervention alone.

Figure 2 illustrates mean diazoxide dose by age at the population level. Longitudinal data on diazoxide course was available for 46 individuals in the medical record cohort (87% of those treated with diazoxide). Diazoxide was discontinued over the follow-up period for one individual in the medical record cohort at the age of 17 years due to improved glycemic control. A statistically significant difference in median age was observed across the three diazoxide dose (mg/kg/day) adjustment groups (p = 0.040, Kruskal-Wallis test). Median age at last recorded diazoxide dose was 6.7 years (IQR, 2.2–11.0 years, n = 14) for individuals who required diazoxide dose increase, 14.1 years (IQR, 12.5–21.0 years, n = 6) for those whose diazoxide dose remained unchanged, and 14.0 years (IQR, 6.3–19.3 years, n = 26) for those whose diazoxide dose was decreased or discontinued. Post-hoc pairwise comparisons using Dunn’s test with Bonferroni correction showed that individuals who required a diazoxide dose increase were statistically significantly younger than those whose diazoxide dose was unchanged (p = 0.041). No statistically significant differences in age were observed between those whose diazoxide dose was increased and those whose dose was decreased (p = 0.053) nor between those whose dose was unchanged and those whose dose was decreased (p = 0.608). Among HIGR participants reporting diazoxide use, two (22%) ultimately discontinued diazoxide; one at age 10–13 years for reasons unknown, and one at age 3–4 years due to adverse effects of treatment.

Fig. 2.

Mean diazoxide dose by age group in the medical record cohort. Filled blue circles depict the diazoxide dose in mg/kg/day for individual patients at each year of age across the follow-up period. Open black circles depict the mean diazoxide dose in mg/kg/day for each year of age. Although a decrease in mean diazoxide dose with advancing age is visually apparent, this analysis is descriptive and reflects cross-sectional aggregation rather than within-individual longitudinal change; therefore, these findings should be interpreted as exploratory.

No individuals in the medical record cohort who had not undergone a pancreatectomy, and none of the HIGR participants with non-missing survey responses (n = 10), developed diabetes mellitus over the follow-up period.

Epilepsy

Twenty-five individuals in the medical record cohort (38%) were diagnosed with epilepsy. Of those diagnosed with epilepsy, 24% had remission of seizures following appropriate hyperinsulinism treatment, 24% had seizures at presentation that persisted despite hyperinsulinism treatment, 28% had seizures at presentation that resolved with hyperinsulinism treatment but then later recurred (median age recurrence: 48 months [IQR, 12–60 months]), and 24% had onset of seizures after initiation of hyperinsulinism treatment (median age onset: 20.5 months [IQR, 16–29 months]). Of the 19 individuals for whom seizures persisted, recurred, or began following initiation of hyperinsulinism treatment, seizures were classified as atypical absence in 47%. Eleven (79%) of the 14 individuals with longitudinal data on antiseizure medication course remained on antiseizure medication over the follow-up period (median age at last follow-up: 14.0 years [IQR, 8.1–26.0 years]). All three individuals for whom antiseizure medication was discontinued had documented seizures off epilepsy treatment. Three HIGR participants (33%, n = 10 responses) reported receiving a diagnosis of epilepsy (defined as recurrent seizures not due to hypoglycemia). All three were diagnosed at 4–9 years of age and all remained on antiseizure medication at the time of HIGR survey completion (age range: 13.8–19.9 years).

Neurodevelopmental Course

History of any neurodevelopmental issue was documented in 64% of the medical record cohort and reported by 73% of HIGR participants (n = 11 responses). Most commonly, these included delayed acquisition of early developmental milestones and attention deficit/hyperactivity disorder (ADHD) (shown in Fig. 3). Median age at diagnosis of ADHD in the medical record cohort was 7.0 years (IQR, 5.9–8.1 years). Among HIGR participants with ADHD, 75% were diagnosed at 4–9 years of age. In the medical record cohort, three individuals (5%) had a documented diagnosis of intellectual disability. History of intellectual disability was not included as a survey response item in HIGR at the time of data extraction; however, no respondents listed intellectual disability as “free text” response. Nearly one-third (27%) of the medical record cohort and two-thirds of HIGR participants (64%, n = 11 responses) had an Individualized Education Plan (IEP) or reported receipt of school accommodations. It was not possible to differentiate whether accommodations were in place to support educational or medical needs. Despite the high prevalence of neurodevelopmental concerns and receipt of school accommodations, only 15% of the medical record cohort and 27% of HIGR participants (n = 11 responses) had documented or reported formal neuropsychological evaluation.

Fig. 3.

Distribution of neurodevelopmental outcomes in the medical record cohort (blue) and HIGR (orange). *n = 8 HIGR responses for questions pertaining to ADHD and n = 11 HIGR responses for the remaining outcomes. ADHD, attention-deficit hyperactivity disorder; ASD, autism spectrum disorder; HIGR, HI Global Registry.

Genotype-Phenotype Associations

In the medical record cohort, patients with pathogenic GLUD1 variants in exons 10, 11, and 12 were younger at presentation (median age: 14.5 days [IQR, 1–101.5] vs. 345 days [IQR, 255–438], p < 0.001, Wilcoxon rank-sum test) and had higher median ammonia levels (136.5 μmol/L [IQR, 91–160] vs. 82.5 μmol/L [IQR, 68–101], p < 0.001, Wilcoxon rank-sum test) than patients with pathogenic GLUD1 variants in exons 6 and 7. Epilepsy that persisted, recurred, or began despite hyperinsulinism treatment was observed in a higher proportion of patients with pathogenic GLUD1 variants in exons 10, 11, and 12 compared to those with variants in exons 6 and 7 (40.5% vs. 14.3%, p = 0.02, χ² test). Differences in neurodevelopmental outcomes between patients with pathogenic variants located in exons 10, 11, and 12 and those with variants in exons 6 and 7 were not observed.

In the pooled analysis including our medical record cohort and previously described cohorts [7, 8, 11], an association between genotype and epilepsy was no longer observed (40.3% exons 10, 11, and 12 vs. 35.6% exons 6 and 7, p = 0.59, χ² test). However, associations between genotype and median age at presentation (60 days [IQR 3–152] vs. 360 days [IQR, 213–382], p < 0.001, Wilcoxon rank-sum test) and median ammonia level (125 μmol/L [IQR, 95–161] vs. 88 μmol/L [IQR, 74–103], p < 0.001, Wilcoxon rank-sum test) persisted.

HIGR: Quality of Life

Parents of individuals with HI/HA reported that they felt that their “life was ruled” (55%) and that they worried (73%) “quite often,” “very often,” or “always” because of their child’s hyperinsulinism or hyperinsulinism-related condition (n = 11 responses). Daily care of hyperinsulinism and hyperinsulinism-related conditions was described as “simple” or “manageable” by 45% of parents and as “demanding,” “complicated,” or “disruptive” by the remaining 55% (n = 11 responses). Parents commonly found the process of obtaining the diagnosis of hyperinsulinism to be “difficult” or “very difficult” (70%, n = 10 responses). Despite these reported challenges, most parents (73%) reported their quality of life as “excellent,” “very good,” or “good” (n = 11 responses).

Discussion

We describe the natural history of HI/HA syndrome utilizing medical record data and survey data from caregivers and individuals with HI/HA via HIGR. In keeping with previous reports, we found that neonates with HI/HA syndrome typically have normal birth weight, and that recognition of hypoglycemia is frequently delayed [7, 11, 17, 18]. Initial presentation most commonly occurred outside of the neonatal period, and not infrequently, was delayed beyond 1 year of age [3, 7, 8, 11, 17, 18]. By contrast, among children with hyperinsulinism due to any cause, 60–70% present within the first week of life [14, 19]. Notably, earlier symptom onset was reported in the HIGR cohort compared to the medical record cohort in our study as well as those previously published. This may reflect a delay between the identification of symptoms by caregivers and the recognition of symptoms by healthcare providers, retrospective identification of symptoms by caregivers (e.g., after receiving the diagnosis of hyperinsulinism, caregivers may recognize that symptoms began earlier than initially reported), recall bias, or bias and limitations in history taking and medical record documentation. In considering these, it is interesting to note that symptoms of hypotonia and poor feeding at presentation were reported by nearly half of HIGR respondents but were infrequently documented in the medical record cohort. In contrast, we found that 70% of individuals in both cohorts in our study had hypoglycemic seizures at presentation. Similarly, published case series describe hypoglycemic seizures as the presenting symptom in 64–88% of individuals with HI/HA syndrome, but hypotonia and feeding issues at presentation were rarely reported [7, 10, 11, 17]. This is particularly striking, because both are well-recognized symptoms of hypoglycemia in infants and suggests that these may be underappreciated early symptoms of hypoglycemia.

The impact of the diagnostic odyssey on caregivers of children with HI/HA syndrome was highlighted by responses to HIGR quality of life surveys. Caregivers of children with HI/HA syndrome reported that the experience of obtaining a diagnosis “difficult” or “very difficult” nearly twice as frequently as caregivers of children with any form of hyperinsulinism [14]. In contrast, responses to questions regarding overall quality of life, worry, and feelings that “life was ruled” by hyperinsulinism among caregivers of children with HI/HA syndrome mirrored those reported by caregivers of individuals with any form of hyperinsulinism [14].

Epilepsy in HI/HA syndrome is common and has been described to occur in approximately 40% of those affected in most studies, with one study reporting a prevalence of 64% [7, 10–12, 17]. We similarly found that approximately 40% of the medical record cohort received a diagnosis of epilepsy. However, nearly one-quarter had complete remission of seizures following appropriate hyperinsulinism treatment reflecting the relatively high frequency of misdiagnosis of epilepsy in both the medical record and HIGR cohorts. Atypical absence seizures, often accompanied by eyelid myoclonus, are a distinctive seizure type observed in HI/HA syndrome. These seizures, which have features of high-amplitude irregular, generalized, spike and wave discharges on electroencephalogram, occur in the setting of euglycemia and are distinct from the focal-onset seizures, seen in other forms of hyperinsulinism, that occur as a consequence of hypoglycemic brain injury [10, 11]. Atypical absence seizures may be easily controlled or refractory to treatment; generalized tonic-clonic seizures and focal seizures have also been described [11, 20]. The types of seizures and age of seizure onset in our cohort were similar to previous reports; however, we were unable to reliably ascertain whether seizures were due to hypoglycemia [10, 11]. We found that the majority of individuals for whom epilepsy persisted or developed following hyperinsulinism treatment continued to require antiseizure medication over the follow-up period. This finding should be interpreted with caution, since longitudinal seizure treatment data were unavailable for one-quarter of the medical record cohort and one-third of HIGR participants did not complete the relevant survey questions on seizure course.

Neurodevelopmental disorders have been reported to be more prevalent in HI/HA syndrome than in other genetic forms of hyperinsulinism [11, 12, 21]. A history of any neurodevelopmental problem was documented in 64% of the medical record cohort and was reported by 73% of HIGR participants. Our findings are similar to those observed in a study of adolescents and adults with HI/HA syndrome in which 75% of participants self-reported a history of neurodevelopmental problems and 25% had history of ADHD [12]. By comparison, it is estimated that 26–48% of children with any form of hyperinsulinism have neurodevelopmental problems and that 11% of all children aged 3–17 years in the USA have ADHD [22]. In particular, a high frequency of intellectual disability has been reported among individuals with HI/HA syndrome. Bahi-Buisson et al. [11] found that 55% of 22 individuals with HI/HA syndrome had an IQ score ≤70 and 14% had an IQ score ≤50. Similarly, Su et al. [7] reported that 42% of 26 individuals with HI/HA syndrome had abnormal scores on either the Gesell Developmental Schedule or Wechsler Intelligence Scale for Children. The low prevalence of intellectual disability observed in our study is in stark contrast to these reports. Reasons for this may include the low rate of documented formal neuropsychological evaluation and ascertainment bias. Notably, the HIGR survey did not include “intellectual disability” or related terms as response items at the time this study was conducted. Over one-third of HIGR participants reported a history of learning disorder and it is possible that this group included individuals with both specific learning disorders and those with intellectual disability.

Hypoglycemia in HI/HA syndrome has been described as typically well-managed with moderate doses of diazoxide [23]. Diazoxide, which inhibits insulin secretion by opening beta-cell K_ATP channels, has a therapeutic dose range of 5–15 mg/kg/day for children and 3–8 mg/kg/day for adults [15]. We observed that, on average, moderate doses of diazoxide were required in infancy and early childhood with low-moderate doses required in late childhood and adolescence. Of note, several individuals were treated with very high diazoxide doses, particularly in the first few years of life. In these cases, supratherapeutic doses of diazoxide were employed prior to inpatient evaluation at the two US HI centers of excellence and were able to be reduced prior to hospital discharge. We suspect that this reflects the profound protein-induced hypoglycemia characteristic of HI/HA syndrome. While diazoxide typically mitigates protein-induced hypoglycemia in HI/HA syndrome, protein-induced hypoglycemia despite treatment with diazoxide, including at high doses, was observed for some within our cohort, and has previously been described [8, 16, 24]. It is thus imperative that affected individuals, including those treated with diazoxide, consume carbohydrates concomitantly with any protein intake. Protein is critical for growth and restricting protein intake below the recommended dietary allowance for age is usually not required.

Improvement in the severity of hypoglycemia, resolution of hyperinsulinism, and subsequent development of diabetes mellitus have been observed in individuals with other genetic forms of hyperinsulinism, including those with pathogenic variants in ABCC8, KCNJ11, HNF1A, and HNF4A treated without pancreatectomy, but have not been well described in HI/HA syndrome [15, 25, 26]. Su et al. [7] reported that diazoxide could be discontinued in 4 (15%) of 26 individuals with HI/HA syndrome and Brandt et al. [27] described resolution of hyperinsulinism at age 8 years in a child with HI/HA syndrome, although details of the biochemical criteria used to demonstrate resolution in both reports were incomplete. In contrast, diazoxide discontinuation was uncommon in the medical record cohort and no individuals demonstrated biochemical resolution of hyperinsulinism. Differential follow-up between individuals with improved versus persistent disease could explain the lower frequency of diazoxide discontinuation and diet-only management in the medical record cohort as compared to that observed in HIGR and previously published cohorts. However, significant protein-induced hypoglycemia has been well documented in adults, including seniors and asymptomatic adults, with HI/HA syndrome, supporting that this is a persistent feature of the disorder [16]. None of the medically managed individuals in the medical record cohort and no HIGR participants developed diabetes mellitus over the follow-up period. A limitation is that few individuals in our cohort were followed into adulthood. Development of type 2 diabetes mellitus has increasingly been anecdotally recognized and has been reported in at least two adults with HI/HA syndrome, highlighting the need for long-term follow-up [16, 28].

A genotype-phenotype association between pathogenic variants in exons 6 and 7 of the GLUD1 gene and epilepsy were initially reported by Bahi-Buisson et al. [11] and corroborated shortly thereafter by Kapoor et al. [8]. However, this association was not observed in subsequent case series or reports. In our medical record cohort, we found an association between pathogenic variants in exons 10, 11, and 12 of the GLUD1 gene and epilepsy. However, in pooled analyses including previously published large cohorts of patients with HI/HA syndrome, these associations were no longer apparent. Interestingly, both in our cohort and in pooled analyses, pathogenic variants in exons 10, 11, and 12 of the GLUD1 gene were associated with younger age at presentation and higher ammonia levels. These findings are exploratory and warrant further investigation.

Our findings need to be interpreted in the context of several limitations. Medical record data were subject to possible incomplete or inconsistent documentation. Additionally, data from other institutions were not universally available. This is particularly relevant for rare disease populations, in which evaluations are frequently distributed between local hospital systems and distant centers of excellence. Since care at our centers was more likely to be hyperinsulinism-related and less likely to include non-endocrine evaluations, this could have biased our findings to underestimate neurologic and neurodevelopmental outcomes. Selection bias was possible for both the medical record and HIGR cohorts. Individuals referred to our institutions, which are both large hyperinsulinism referral centers, and who elected to complete HIGR surveys may have had more severe or complex disease, leading to overestimation of outcomes. Conversely, our cohort may have differentially included individuals from more advantaged backgrounds, including those with high parental involvement and enhanced access to specialized services, and thus potentially better outcomes. Consequently, our findings may not be generalizable or representative of the broader population of patients with HI/HA syndrome.

Key strengths of this study include the relatively large cohort size, detailed phenotypic and molecular characterization of affected individuals in the medical cohort, and the unique incorporation of patient-reported registry data, which provided complementary information on outcomes and perspectives often underrepresented in clinical records. That findings in both the medical record cohort and the HIGR cohort largely mirrored previously reported findings from smaller case series supports the validity of our conclusions and highlights the potential value of patient-driven data in addressing existing knowledge gaps.

Conclusion

Recognition of hypoglycemia in patients with HI/HA syndrome is often delayed. Presentation with hypoglycemic seizures and initial misdiagnosis with epilepsy are common. While hypoglycemic seizures resolve with treatment, epilepsy often recurs in childhood. Longitudinal neuropsychological assessment by specialists familiar with HI/HA syndrome is critical because the characteristic seizure type observed in HI/HA syndrome, atypical absence seizures, may be subtle and neurodevelopmental issues are common. Among those who develop epilepsy, the need for epilepsy treatment appears to persist into adulthood, although data on longitudinal seizure course in HI/HA syndrome remain limited. Hypoglycemia in HI/HA syndrome is treatable with diazoxide. Diazoxide requirement tends to decrease with age; however, most patients continue to require diazoxide requirement into adulthood reflecting the profound protein-induced hypoglycemia observed in HI/HA syndrome. Although HI/HA syndrome has been well characterized in the nearly 3 decades since its molecular etiology was uncovered, persistent gaps in our understanding of its natural history remain. Prospective longitudinal studies incorporating patient-centered data are needed.

Acknowledgments

The authors would like to thank Nicole Wittmeyer, MPH, and Melinda Danowitz, MD, for their help with data extraction. Additionally, we would like to thank HIGR participants and their caregivers for their contribution to HIGR.

Statement of Ethics

The protocol for this study was approved by the Children’s Hospital of Philadelphia Institutional Review Board (IRB), Protocol No. 22-019652; Cook Children’s Medical Center IRB, Protocol No. 2022-005; and North Star Review Board IRB (which reviewed and approved the HIGR research study), Protocol No. NB100012, respectively. All HIGR study participants provided informed consent to participate in research when they joined the HI Global Registry. The Children’s Hospital of Philadelphia IRB and Cook Children’s Medical Center IRB granted exemption from requiring written informed consent for medical record data abstraction for this study, respectively.

Conflict of Interest Statement

D.D.D.L. was a member of the journal’s Editorial Board at the time of submission. D.D.D.L. has received consulting fees from Zealand Pharma A/S, Rezolute, Rhythm Pharmaceuticals, Confo Therapeutics, AmideBio, Spruce Biosciences, Ligand Pharmaceuticals, Twist Biosciences, and Fortress Biotech. D.D.D.L. has received research funding from Hanmi Pharmaceutical, Twist Biosciences, Zealand Pharma A/S, Rezolute, Ultragenyx, Rhythm Pharmaceuticals, and Moderna for studies not related to this manuscript. T.L.S.P. and J.R. are employees of CHI. Over the past 3 years, CHI has received sponsorships for events and research and support programs from Hanmi Pharmaceutical, Rezolute, Rhythm Pharmaceuticals, and Zealand Pharma. P.S.T. received payments independent of this work from Ascendis, Novo Nordisk, Rezolute, Spruce Biosciences, Xeris Pharmaceuticals, and Zealand Pharma A/S for performing research studies and consulting fees from Nutrition North America. His institution has received payments on his behalf for consulting from Neurocrine Biosciences, Spruce Biosciences, Crinetics, Rezolute, and Zealand Pharma A/S. The remaining authors have no conflicts of interest to declare.

Funding Sources

This work was supported by grants from the University of Pennsylvania Orphan Disease Center in partnership with Team CHIbra and CHI (MDBR-22-109-CHI) and by the National Institutes of Health (K23DK136967 [E.R.]). Hanmi Pharmaceutical, Rezolute, and Zealand Pharma are CHI financial sponsors, whose sponsorship helps support HIGR. CHI also receives funds for awareness and HIGR through the LightCure grant funded by the European Union (101080327). CHI has many individual and foundation donors funding other CHI projects and infrastructure. The funders had no role in the design, data collection, data analysis, and reporting of this study.

Author Contributions

E.R. designed the study, acquired and analyzed data, and wrote the manuscript. L.M.M., K.E.B., T.L.S.P., J.R., and D.R. acquired data and critically revised the manuscript. O.T. analyzed data and critically revised the manuscript. P.S.T. and D.D.D.L. conceptualized the work and critically revised the manuscript.

Funding Statement

This work was supported by grants from the University of Pennsylvania Orphan Disease Center in partnership with Team CHIbra and CHI (MDBR-22-109-CHI) and by the National Institutes of Health (K23DK136967 [E.R.]). Hanmi Pharmaceutical, Rezolute, and Zealand Pharma are CHI financial sponsors, whose sponsorship helps support HIGR. CHI also receives funds for awareness and HIGR through the LightCure grant funded by the European Union (101080327). CHI has many individual and foundation donors funding other CHI projects and infrastructure. The funders had no role in the design, data collection, data analysis, and reporting of this study.

Data Availability Statement

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of participants but are available from the corresponding author (E.R., rosenfelde@chop.edu) upon request.