Article Text

Download PDFPDF

Safety and efficacy of low-dose diazoxide in small-for-gestational-age infants with hyperinsulinaemic hypoglycaemia
  1. Suresh Chandran1,2,3,
  2. Pravin R R4,
  3. Chua Mei Chien1,2,3,
  4. Seyed Ehsan Saffari5,
  5. Victor Samuel Rajadurai1,2,3,
  6. Fabian Yap2,3,4
  1. 1 Department of Neonatology, KK Women's and Children's Hospital, Singapore
  2. 2 Paediatrics Academic Clinical Programme, Duke-NUS Medical School, Singapore
  3. 3 Paediatrics Academic Clinical Programme, Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore
  4. 4 Department of Pediatrics, KK Women's and Children's Hospital, Singapore
  5. 5 Center for Quantitative Medicine, Duke-NUS Graduate Medical School, Singapore
  1. Correspondence to Professor Suresh Chandran, Neonatology, KK Women's and Children's Hospital, Singapore, Singapore, Singapore; profschandran2019{at}; Dr Fabian Yap; fabian.yap.k.p{at}


Objectives Diazoxide (DZX) is the drug of choice for treating hyperinsulinaemic hypoglycaemia (HH), and it has potentially serious adverse effects. We studied the safety and efficacy of low-dose DZX in small-for-gestational-age (SGA) infants with HH.

Design An observational cohort study from 1 September 2014 to 31 September 2020.

Setting A tertiary Women’s and Children’s Hospital in Singapore.

Patients All SGA infants with HH.

Intervention Diazoxide, at 3–5 mg/kg/day.

Main outcome measures Short-term outcomes; adverse drug events and fasting studies to determine ‘safe to go home’ and ‘resolution’ of HH.

Results Among 71 836 live births, 11 493 (16%) were SGA. Fifty-six (0.5%) SGA infants with HH were identified, of which 27 (47%) with a mean gestational age of 36.4±2 weeks and birth weight of 1942±356 g required DZX treatment. Diazoxide was initiated at 3 mg/kg/day at a median age of 10 days. The mean effective dose was 4.6±2.2 mg/kg/day, with 24/27 (89%) receiving 3–5 mg/kg/day. Generalised hypertrichosis occurred in 2 (7.4%) and fluid retention in 1 (3.7%) infant. A fasting study was performed before home while on DZX in 26/27 (96%) cases. Diazoxide was discontinued at a median age of 63 days (9–198 days), and resolution of HH was confirmed in 26/27 (96%) infants on passing a fasting study.

Conclusion Our study demonstrates that low-dose DZX effectively treats SGA infants with HH as measured by fasting studies. Although the safety profile was excellent, minimal adverse events were still observed with DZX, even at low doses.

  • endocrinology
  • neonatology
  • pharmacology
  • therapeutics

Data availability statement

Data are available on reasonable request. The data that support the findings of this study are available from the corresponding author on reasonable request.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

What is already known on this topic?

  • Small-for-gestational-age infants are at risk of hyperinsulinaemic hypoglycaemia and hypoglycaemic neuronal injury, which has a long-term neurodevelopmental impact.

  • The dose of diazoxide to treat hyperinsulinaemic hypoglycaemia ranges from 5 to 20 mg/kg/day, even in small-for-gestational-age infants.

  • Conventional doses of diazoxide are associated with potentially serious adverse events, including pericardial effusion and necrotising enterocolitis, especially in small-for-gestational-age infants.

What this study adds?

  • Low-dose diazoxide at 3–5 mg/kg/day effectively treats hyperinsulinaemic hypoglycaemia in small-for-gestational-age infants.

  • Although the safety profile was excellent, minimal adverse events were still observed with low-dose diazoxide.

  • Using standardised dilution guidelines to keep the osmolality of the diazoxide suspension <450 mOsm/kg further minimised the risk of drug-induced gastrointestinal complications.


Hypoglycaemia occurs in neonates due to variations in metabolic adaptation during the establishment of enteral feeding in early life. Brain injury with permanent cognitive disability following refractory hypoglycaemia has been reported.1 The infant at risk of hypoglycaemia includes small for gestational age (SGA) and large for gestational age, preterm and infants of diabetic mothers.2 The aetiology of hypoglycaemia in SGA infants includes reduced hepatic glycogen stores, defective gluconeogenesis and hyperinsulinism.3 Perinatal stress hyperinsulinism (PSHI) is now an accepted term for hyperinsulinaemic hypoglycaemia (HH) in SGA infants, and the natural history involves the resolution of hyperinsulinism within 6 months.4 5 Infant at risk of hypoglycaemia needing a glucose infusion rate (GIR) >10 mg/kg/min to maintain normoglycaemia (3.5–7 mmol/L) after 48 hours of life raises suspicion of HH.2 Diazoxide (DZX) is the only drug approved as first-line therapy for HH by the Food and Drug Administration, USA.6 DZX mediates its action through the sulfonylurea receptor-1 subunit of the KATP channel of pancreatic β cells, preventing insulin release. An increasing number of serious adverse effects related to DZX, such as pulmonary hypertension (PH) and necrotising enterocolitis (NEC), have been reported.7 Despite its prevalent use in HH, there is a paucity of literature on the use of low-dose (<5 mg/kg/day) DZX in SGA infants. Our study aimed to characterise the short-term safety and efficacy of low-dose DZX in SGA infants with HH.

Material and methods

Study setting

This observational cohort study was conducted in Singapore at a tertiary hospital between 1 September 2014 and 31 September 2020. Our institution is the largest tertiary perinatal referral centre in the country and conducts approximately 11 000+ deliveries per year. We have a 40-bed level 4 neonatal intensive care unit and a 60-bed special care unit to serve high-risk infants.

Patient population and definitions

We included all SGA infants with HH in this study. The cohort consisted of 56 preterm and term SGA infants with HH. Small-for-gestational-age infants are defined as birth weight for gestational age <10th percentile based on a sex-specific reference population.8

During the fetal–neonatal transition, glucose levels drop to a mean nadir of 2.8 mmol/L by 1 hour of age and rise with feeds and counter-regulatory responses to mean levels of >3.3 mmol/L by 2 hours.9 Based on recommendations from the Pediatric Endocrine Society,10 we target capillary (whole) blood glucose (CBG) levels of >3 mmol/L and >3.5 mmol/L before and after 48 hours of life, respectively, for at-risk infants. Symptomatic hypoglycaemia was considered when any of the following was observed: jitteriness, seizures, apnoea, cyanosis, poor feeding, irritability, lethargy or hypothermia.2

Hypoglycaemia screening

All SGA infants undergo hypoglycaemia screening from birth. Management includes early breast feeding, skin-to-skin care, CBG testing and the use of oral glucose gel (OGG), Rapilose (Penlan Healthcare Ltd, UK). A CBG level is checked at 2 hours of age unless the infant was symptomatic as the physiological glucose nadir occurs in the first hour of life.9 Subsequent CBG checks are at 6, 12, 18, 24 and 36 hours of life. As infants need metabolic substrate during placental to enteral transition and SGA infants have insufficient glycogen stores, early feeding ensures substrate availability.

From 2 hours of age, asymptomatic infants with CBG <3 mmol/L were managed with 0.5 mL/kg of OGG applied to the buccal mucosa, followed by a milk feed. If they remained hypoglycaemic but showed a rising trend of CBG, a repeat dose of OGG and extra feeds are offered before resorting to intravenous dextrose.

Hyperinsulinism: definition and diagnosis

Hyperinsulinaemic hypoglycaemia is suspected when an infant >48 hours of life requires GIR >10 mg/kg/min. We confirmed HH when hypoglycaemia occurred in the presence of detectable serum insulin (>1.6 mU/L), low blood ketones (<0.6 mmol/L) and low fatty acid levels (<0.5 mmol/L).11

Diazoxide treatment

We commenced DZX if there is:

  1. Recurrent need to increase GIR to maintain normoglycaemia after 48 hours of age.

  2. Recurrent hypoglycaemic episodes despite increasing feeds.

  3. Failure to achieve full feeds while weaning intravenous glucose.

This approach recognises the possibility that spontaneous resolution of HH may occur with gradual reduction of GIR (not >1 mg/kg/min) and graded escalation of feed volume (to 180 mL/kg/day) using fortified milk while monitoring CBG.

Liver and renal function tests are routinely performed before initiation of DZX since it is metabolised in the liver and excreted by kidneys. A 2D-echocardiographic evaluation is needed as DZX may cause PH and fluid retention, worsening pre-existing pericardial effusion (PE).7

Enteral preparation of DZX includes an oral suspension Proglycem (Teva Pharmaceuticals, USA) or a compounded oral liquid formulation. The latter is cheaper but has issues with stability and short expiry.12 We use the ready-made DZX suspension. To reduce feed intolerance and risk of NEC, a standardised dilution guideline for DZX to maintain the osmolality <450 mOsm/kg is used for infants weighing <1800 g and aged ≤35 gestational weeks.13 14 Our starting dose of DZX for SGA infants is 3 mg/kg/day in two divided doses with hydrochlorothiazide 1 mg/kg/day in two divided doses to reduce fluid retention. Further increments of DZX are done at a rate of 2–2.5 mg/kg/day if the response is inadequate after 72 hours. Adverse effects of DZX therapy include fluid retention, hypertrichosis, PE, neutropenia, feed intolerance, NEC and PH.7

Home care plans are initiated once CBG is stable on DZX and full feeds are achieved. Caregiver training includes home monitoring of CBG and training on hypoglycaemia and hyperglycaemia contingency plans. Parents are given an outline of side effects and measures to observe. In addition, a safety fasting study (SFS) is performed to determine a safe fasting interval (6–8 hours in <6 months old)15 in case of an inadvertent fast at home.

On follow-up, we review the home glucose profile, weight gain, blood counts and serum electrolytes. An echocardiogram is repeated if the infant developed side effects while on treatment or if oedematous.

Self-weaning of DZX continues with weight gain. We actively reduced the dose of DZX if the home glucose level is persistently >7 mmol/L. When the dose of DZX reaches ≤1.5 mg/kg/day with normal CBG, we stop DZX for 3 days while CBG monitoring continues at home. The infant is then admitted to the hospital for a supervised resolution fasting study (RFS).15

Safety and effectiveness evaluation criteria

Adverse drug reactions (ADRs), defined as an adverse event for which a causal relationship to DZX could not be excluded, were used to measure safety. Fasting studies were used as measures for effectiveness. The SFS determined when it was ‘safe to go home’ on DZX, while the RFS determined the ‘resolution’ of HH without DZX.

Data collated include patient demographics, GIR, date of DZX initiation, dose, response to treatment, ADR and fasting study outcomes.

Statistical analysis

Data were analysed with SAS software V.9.4 for Windows (Cary, NC: SAS Institute Inc.). Demographics and clinical features were reported as descriptive statistics via frequency and percentage for categorical variables and mean±SD and median (minimum–maximum) for continuous variables. Dose of DZX was reported for the term and preterm SGA groups. Treatment data (day of initiation, DZX dose, duration of treatment) were presented using the mean±SD and median (minimum–maximum).


Among 71 836 live births, 11 493 (16%) were SGA, of which 56 (0.5%) had HH. Twenty-seven (47%) SGA-HH infants required DZX treatment, while the rest spontaneously resolved with glucose infusion and feeds. Fifty-two per cent of DZX-treated infants had PSHI indicated by metabolic acidosis (cord blood) and/or abnormal fetal CTG, but none had a 5 min Apgar <6, while maternal hypertension was present in 33%. Among these, 15/27 (55%) were term with male predominance (55%). Clinical characteristics of DZX-treated infants are presented in table 1. Of note, five (18.5%) infants had jitteriness, and one (3.7%) had seizures, and they presented at a mean age of 1.1±0.4 days. All had normal plasma ammonia levels. Diagnosis of HH was made at a median age of 6 days. Diazoxide was initiated at a median age of 10 days. Mean effective dose of DZX required was 4.6±2.2 mg/kg/day, with 89% receiving 3–5 mg/kg/day (table 2). The effectiveness of low-dose DZX is demonstrated in figure 1. Glucose control while on full feeds and DZX was achieved at a median age of 12 days, and the median age at discharge was 21.5 days, which were statistically comparable with the respective median ages of 16 and 22 days in the spontaneous resolution group. Median duration of central venous line requirement was 14 days. Median duration of DZX treatment until HH resolution was 63 days (table 3).

Figure 1

Before-and-after graph showing the day of life when diazoxide was initiated (DZX start) to the day of life when hyperinsulinaemic hypoglycaemia was controlled on diazoxide as measured by safety fasting study (HH end).

Table 1

Demographic and clinical characteristics of diazoxide-treated infants

Table 2

Dose of diazoxide used in term and preterm SGA infants

Table 3

Data of diazoxide treatment

Two infants (7.4%) on DZX 10 mg/kg/day developed generalised hypertrichosis, while others had minimal localised hypertrichosis. One infant (3.7%) who commenced DZX from day 5 developed tachypnoea, PE and hyponatremia on day 9 while on a dose of 4.8 mg/kg/day. Symptoms resolved with fluid restriction, diuretics and discontinuation of DZX. Coincidentally, glucose levels stabilised on full feeds, and she passed RFS before discharge. All other infants, 26/27 (96%), passed SFS before discharge. On stopping DZX, 26/27 (96%) passed a formal RFS (one returned home abroad).

Overall, 24/27 (89%) infants had resolution of HH with low-dose DZX. However, one infant on low-dose DZX had fluid retention.


The incidence of HH among SGA infants is rising worldwide, especially in Southeast Asia.16 17 Our data demonstrated that the combination of HH and SGA occurs in 1:1282 live births, which contrasts with the reported incidence of 1:12 000 live births by Hoe et al.5 Consequently, DZX use is more prevalent as more SGA infants with HH are diagnosed.18 Among our prospectively studied SGA live births, the incidence of HH was 1:205, unlike the estimate of 5:27 by Collins et al.3 To the best of our knowledge, this is the first report that depicts the safety and efficacy of low-dose DZX to treat HH in SGA infants. Our study shows that low-dose DZX has excellent efficacy in SGA infants with HH and minimal ADR.

The aetiology of HH in SGA infants is multifactorial, and a genetic aetiology is unlikely.19 An epigenetic explanation involving hypoxia-inducible factor 1 in SGA is a possible mechanism, which can transiently alter the gene expression causing β-cell dysregulation resulting in HH.20 21

Hypoglycaemia in SGA infants is known to occur at a median age of 1 (0–168) day.5 This was also observed in our cohort, where 22% were symptomatic. Hoe et al confirmed HH at a median age of day 13,5 similar to day 10 in our cohort. We commenced DZX at a median age of 10 days, consistent with the recent recommendations by Brar et al to allow 7–10 days for spontaneous resolution.22

Previous key publications indicate that the effective DZX dose ranged between 5 and 20 mg/kg/day, even for SGA infants.5 19 Many centres initiate DZX at 10 mg/kg/day, but some begin at 5 mg/kg/day in SGA infants.23 Hoe et al and Raisingani and Brar indicate that the dose of DZX required for optimal glucose control in SGA-HH infants was 8 and 10 mg/kg/day, respectively.5 24 In contrast, 89.1% of our SGA cohort required a DZX dose of 3–5 mg/kg/day. The steady-state concentration of DZX allows a twice-daily dosing schedule.25 For safety, all infants in our cohort underwent an assessment of liver and kidney function before DZX initiation.

Biochemical screening includes cortisol and growth hormone during workup for HH. SGA infants may not mount an appropriate cortisol response to hypoglycaemia, and HH can blunt cortisol response,26 warranting ACTH stimulation testing. In our cohort, five (18%) had clinically concerning cortisol levels (<100 nmol/L) and required stimulation testing, which all of them passed.

The median duration of DZX therapy in our population was 63 days, substantially lower than the 181 days reported by Hoe et al,5 who like us, determined resolution using fasting studies. In contrast, our duration of therapy was similar to the 52 days reported by Raisingani and Brar,24 although their DZX dose was two to three times higher.

Feed intolerance and NEC has been increasingly reported in infants receiving oral DZX.22 27 Recently, NEC was reported in preterm infants following DZX therapy for HH.27 In our cohort, feeding-related issues while on DZX were not observed, and this can be attributed to the use of low-dose DZX, palatability of readymade suspension, use of human milk feeds and optimisation of the osmolality of DZX suspension.13 14 Fukutomi et al reported hypertrichosis in 8.6% of Japanese children treated with DZX and resolution in 11 months on discontinuation.28 In our cohort, generalised hypertrichosis was limited to those on a dose of 10 mg/kg/day, which could indicate dose dependency.29

Oedema has been reported with DZX use in 18% of cases.30 Re-opening of the ductus arteriosus and heart failure due to fluid retention and worsening PH with higher doses of DZX were reported.19 22 Reports of large symptomatic PE while on long-term high-dose DZX warrant close cardiac monitoring.31 However, infants who developed ADR often had confounding risk factors like extreme prematurity, congenital heart disease or hypoxic–ischaemic encephalopathy.18 31 One of our SGA infants developed fluid retention, which resolved when DZX was discontinued. A meta-analysis reported PH in 2.4% of DZX-treated infants with HH,30 but none in our cohort. Thornton et al reported PH in 7.6% of SGA infants who had PSHI compared with 1.2% with genetic HH, while on DZX.32 Cerebral insufficiency following intractable hypotension and periventricular leukomalacia was reported in a preterm SGA infant with HH treated with 9 mg/kg/day of DZX.33

Strengths and limitations

We identified 16% of babies as SGA, although we expected 10%. Singapore birth norms superimposed well with Fenton 2013 charts between 25 and 39 weeks’ gestation and then separated from 39 to 42 weeks,34 implying that overestimation of SGA birth size may have occurred in full-term babies. In our cohort, none were lost to follow-up except one who returned abroad. Safety protocols before starting DZX were strictly maintained, including performing echocardiograms. Our study highlights the importance of fasting studies in the management of HH as it determined the endpoints. Safety fast study assures safety before discharge on DZX while confirming the effectiveness of DZX dose. On the other hand, the RFS is required as it formally establishes that HH has resolved.5 15

While our SGA-HH cohort size was modest, it is comparable with Hoe et al and Raisingani and Brar.5 24 Although our study was conducted in a large Women’s and Children’s hospital with general and tertiary services, larger multicentre studies are required given the low incidence of HH in SGA. Since this is a short-term study, the long-term outcome of infants on DZX needs to be studied, as DZX hyperpolarises the membrane potential of pancreatic and also of the brain and intestinal cells, increasing the risk of neuronal injury and NEC.35 36


Low-dose DZX is effective in treating SGA infants with HH. Although the safety profile was excellent, adverse events were observed with DZX, even at low doses. We recommend commencing DZX at 3 mg/kg/day and titrating upwards in SGA-HH infants as, in our experience, such an approach provides an optimal risk–benefit balance.

Data availability statement

Data are available on reasonable request. The data that support the findings of this study are available from the corresponding author on reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

This study was exempted from ethics approval by the SingHealth Centralized Institutional Review Board (CIRB 2021/2166).



  • SC and FY contributed equally.

  • Contributors SC and FY conceptualised and designed the study, wrote the manuscript and performed the final edits. Both authors supervised the treatment of patients and coordinated the hyperinsulinism service. PRR drafted the initial manuscript and inserted the references. CMC and VSR reviewed and revised the manuscript critically for important intellectual content. SES is the biostatistician who is responsible for the data analysis and interpretation. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.