Context: Congenital hyperinsulinism (HI) is a serious clinical feature of Beckwith-Wiedemann syndrome (BWS) causing severe hypoglycemia. The relationship between BWS genotypes and HI severity is not well understood.

Objective: Investigate the relationship between molecular determinants of patients with BWS and HI with measures of HI severity.

Design: Retrospective cohort study including 85 children from 2009-2024.

Setting: All patients evaluated at single, tertiary care center.

Patients: BWS genotype frequency included 41 children with pUPD11, 24 with IC2 LOM, eight with 11p15 chromosomal anomalies, six with GWpUPD, four with IC1 GOM, and two with CDKN1C.

Interventions: Retrospectively reviewed interventions included maximum glucose infusion rate (max GIR), diazoxide responsiveness, and surgery.

Main Outcome Measures: Primary outcome was association between BWS genotypes and measures of HI severity. Secondary outcomes included the relationship between pUPD11 length and presence of K-ATP variants with diazoxide responsiveness and surgical need.

Results: Significant differences presented among genotypes in max GIR (p = 0.004), enteral dextrose requirements (p = 0.029), and pancreatectomy (p = 0.012). Most patients with IC2 LOM, IC1 GOM or CDKN1C were diazoxide responsive and did not require surgery. Patients with pUPD11 were more likely to be diazoxide unresponsive and require surgery, especially if pUPD11 length extended into the K-ATP gene region and if a pathogenic variant in the ABCC8 or KCNJ11 was present.

Conclusion: Patients with pUPD11 experience more severe HI, while patients with IC2 LOM, IC1 GOM, and CDKN1C exhibit milder disease. Based on our findings, we designed a genetic testing algorithm to guide clinical management.

Patient Cohort

All patients included in this study received care at the Children’s Hospital of Philadelphia Congenital Hyperinsulinism Center. A retrospective chart review was conducted on medical records of patients with molecularly confirmed BWS and a diagnosis of hyperinsulinism between 2009 and 2024. Among all patients with BWS cared for at the Children’s Hospital of Philadelphia between 2009 to 2024, only those with a molecularly confirmed diagnosis were eligible for inclusion and patients with only a clinical diagnosis of BWS were excluded. From this molecularly confirmed BWS cohort, patients were included if they also had a diagnosis of hyperinsulinism. The diagnosis of HI was established when hypoglycemia persisted beyond the first week of life and was supported by additional criteria such as increased glucose requirement, inappropriately low plasma concentrations of beta-hydroxybutyrate and free fatty acids, and a glycemic response to glucagon. Molecular testing via single nucleotide polymorphism (SNP) array and chromosome 11p15 methylation analysis was performed on various tissue types including blood, skin, pancreas, and/or liver. Each patient’s molecular subtype was classified as IC2 LOM, pUPD11, IC1 GOM, CDKN1C, GWpUPD, or an 11p15 chromosomal anomaly (Table 1). Patients with 11p15 chromosomal anomalies had either a deletion, duplication, partial trisomy, or translocation affecting the 11p15 region. Methylation levels at IC1 and IC2 were assessed using allele-specific methylated multiplex quantitative real-time PCR, with the proportion of cells with exhibiting altered methylation calculated as previously described. Methylation testing results on blood were reviewed, and the percentage of mosaicism was calculated using established methodology^ ^(Supplementary Table 1). Molecular testing for hyperinsulinism focused on Sanger sequencing of ABCC8, KCNJ11, and GCK which accounted for all pathogenic variants identified in this cohort, despite expansion of additional genes to hyperinsulinism panels as the study period progressed. BWS clinical scores were calculated for each patient according to the BWS international consensus guidelines. BWS cardinal and suggestive features are also listed in Supplementary Table 2. Continuous enteral dextrose support via gastrostomy tube to maintain euglycemia was administered alone or in combination with other therapies to allow for at-home management in patients who could not be weaned from glucose support in the hospital, and the total number of patients was captured for analysis (Table 2). Diazoxide responsiveness was defined as the ability to maintain plasma glucose concentrations >70 mg/dL (3.9 mmol/L), after an adequate dose of diazoxide (ranging from 5 – 15 mg/kg/day depending on the patient’s age and divided into two doses), for at least 12 hours of fasting and/or generate appropriate hyperketonemia (plasma beta-hydroxybutyrate >1.8 mmol/L) before development of plasma glucose <50-60 mg/dL (2.8-3.3 mmol/L). Chlorothiazide, a diuretic, was commonly used concurrently with diazoxide to mitigate side effects such as fluid retention. All clinical data was abstracted from medical charts in adherence with IRB 19-016549 and IRB 13-010658.

Statistical Analysis

Statistical analyses were performed using R v4.4.2. Associations between the BWS molecular subtype and continuous variables including maximum glucose infusion rate (GIR), percentage requiring enteral dextrose, frequency of pancreatectomy, BWS clinical score, and blood mosaicism percentage were examined using Kruskal-Wallis or chi-square tests. Potential outliers were assessed using Grubbs’ test and the interquartile range (Tukey) method. All extreme values were verified against source records.

Logistic regression analysis was conducted to examine factors influencing diazoxide responsiveness and the need for pancreatectomy in this patient population. Firth’s logistic regression, which applies penalized maximum likelihood estimation to mitigate bias in rare disease data and unbalanced sample distributions, was used to model likelihood of diazoxide responsiveness and the need for pancreatectomy. To refine analysis, the pUPD11 subtype was further stratified into three groups (Figure 1):

1.     pUPD11-LOH/KATP: Patients with the pUPD11 subtype with extended loss of heterozygosity encompassing the ABCC8 & KCNJ11 gene region and with a confirmed pathogenic variant in ABCC8 or KCNJ11.

2.     pUPD11-LOH: Patients with the pUPD11 subtype exhibiting extended LOH encompassing the ABCC8 & KCNJ11 gene region, but without a detected pathogenic ABCC8 or KCNJ11 variant.

3.     pUPD11: Patient with the pUPD11 subtype in which the LOH does not extend into the ABCC8 or KCNJ11 gene region and without a pathogenic ABCC8 or KCNJ11 variant.

The stratification of pUPD11 was made to compare the effects of each molecular determinant on specific outcomes of hyperinsulinism severity. Patients in the IC2 LOM group were chosen as the reference group due to the sizeable cohort and their predominantly diazoxide-responsive nature, with some variation. SNP arrays, performed at our institution, were analyzed to confirm the length of heterozygosity loss in each patient with pUPD11. Pre-specified exclusion criteria comprised the exclusion of patients with a diagnosis of 11p15 chromosomal anomaly and GWpUPD from this regression analysis due to the increased complexity of these subtypes, which made meaningful comparisons challenging. For the diazoxide responsiveness analysis, patients who had not undergone a diazoxide trial or had their treatment with diazoxide discontinued due to side effects, were excluded.