Scenario

A 33^4/7 weeks’ gestation male neonate, small for gestational age secondary to placental insufficiency (birth weight 1309 g, second centile) is admitted to the neonatal intensive care unit (NICU) for prematurity and mild respiratory distress syndrome. During the first few days of admission, the patient has severe hypoglycaemia that is not responsive to enteral or intravenous interventions. Despite significant escalation in glucose infusion rate (GIR) up to 18 mcg/kg/min, the infant has ongoing episodes of hypoglycaemia; thus, a glucagon infusion is started on day 2 and increased to a peak dose of 0.02 mg/kg/hour. On day 9, the baby is stable in room air and has been able to be weaned off glucagon, but still requires parenteral nutrition with a high GIR, via a peripherally inserted central line. High insulin levels (60–65 pmol/L) were detected during two different hypoglycaemic episodes, and the rest of the critical labs were normal. Feeds are increased with good tolerance; in order to optimise glucose levels, continuous feeds are initiated and fortification is escalated to 0.85 kcal/mL to increase GIR. Despite these interventions, plasma glucose levels are still between 3.2 mmol/L and 3.3 mmol/L. Paediatric endocrinology was consulted; diazoxide treatment is suggested as an option to wean intravenous GIR and progress to full feeds, but given the baseline risk factors of prematurity and intrauterine growth restriction, the possible risk of necrotising enterocolitis (NEC) is brought up as a possible factor influencing the treatment decision.

Structured clinical question

What is the incidence of NEC in neonates treated with diazoxide?

Search strategy and outcome

We searched PubMed, Embase, Cochrane CENTRAL, Scopus and CINAHL for studies evaluating use of diazoxide in both term and preterm neonates diagnosed with persistent refractory hypoglycaemia, defined as hypoglycaemia requiring high GIR (GIR >12 mg/kg/min) beyond 72 hours of age, at a postnatal age of <28 days (for term newborns) or <42 weeks’ corrected gestational age (for preterm infants). The primary outcome was the incidence of NEC, any stage, based on modified Bell’s criteria,1 during NICU admission. Analysis was carried out with software R V.4.1.0 and package ‘meta’ (R Foundation for Statistical Computing, Vienna, Austria), using a random effects model. Heterogeneity of studies was assessed using the I² value, with values interpreted following Cochrane Handbook guidelines: 0%–40%, might not be important; 30%–60%, may represent moderate heterogeneity; 50%–90%, may represent substantial heterogeneity; and 75%–100%, considerable heterogeneity.

A total of seven observational studies were included (n=1410),2–8 and the summary of included studies is presented in table 1. Pooled analyses from seven observational studies showed that the incidence of NEC among neonates with hypoglycaemia treated with diazoxide was 4.4%, with the 95% CI being 1.4% to 12.9% and the prediction interval (interval in which a new similar study may probably fall if selected from the same population) being 0.09%–70%. Results showed substantial heterogeneity (I²=90%).

Commentary

In our review, we found seven observational studies in neonates receiving treatment with diazoxide for hyperinsulinaemic hypoglycaemia that reported rates of NEC. We found an overall incidence of 4.4%, which is concerning in the population included in these studies, whose median gestational ages ranged from 33 weeks to 37 weeks. In moderate to late preterm neonates, NEC incidence has been reported to be 0.1%–0.7%,9 10 and in our pooled results even the lower margin of the 95% CI (1.4%) is significantly higher than what has been described in literature. This suggests a marked increase in development of NEC that should make clinicians aware and cautious when starting diazoxide treatment, especially in high-risk infants. However, the 95% CI is wide (1.4% to 12.9%), which makes it uncertain what is the true value of incidence in this group of infants, and the prediction interval is even wider (0.09%–70%), which means that future studies in different settings may find individual values different from the ones we systematically review in this article.

Neonatal hypoglycaemia is common among both term and preterm infants admitted to the NICU,11 and more than 50% of neonates born with risk factors such as small or large for gestational age, maternal diabetes or prematurity will have at least one hypoglycaemic episode during admission.11 Neonatal definitions for hypoglycaemia differ, and the current Canadian Paediatric Society statement defines the threshold for treatment and investigation as plasma glucose of ≤2.6 mmol/L (47 mg/dL) in the first 72 hours after birth and ≤3.3 mmol/L (59 mg/dL) thereafter.12 Persistent hypoglycaemia has been reported as an independent risk factor for adverse neurodevelopmental outcomes,13 and in this population is most frequently caused by hyperinsulinism, which can be secondary to maternal gestational diabetes, small for gestational age, perinatal stress and less commonly secondary to genetic abnormalities.14 The diagnosis of hyperinsulinaemic hypoglycaemia is typically based on clinical presentation and laboratory markers consisting of low ketones, low fatty acids and detectable insulin during a hypoglycaemic episode, although the latter might not always be present, given short half-life and pulsatile secretion of insulin.15 Management of hyperinsulinaemic hypoglycaemia includes increasing the glucose intake both enterally and intravenously, administering nasogastric feeds over a longer period of time, starting continuous enteral feeds, and prescribing pharmacological measures. This last group includes glucagon, steroids, octreotide and, most commonly, diazoxide.14 16

Diazoxide is a benzothiazine derivative that inhibits depolarisation of pancreatic beta cells, reducing insulin secretion and thus improving glucose levels.17 It has been shown to be effective in paediatric18 and neonatal19 cases of hyperinsulinaemic hypoglycaemia. The safety profile of diazoxide has been extensively studied in paediatric patients with hyperinsulinism,20 but there is a significant gap in the literature about safety of diazoxide in neonatal patients. Neonatal patients with hyperinsulinaemic hypoglycaemia, often premature or growth restricted, often have a fragile cardiovascular physiology that puts them at risk of diazoxide-associated fluid retention and pulmonary hypertension that have been reported in previous literature,21 with a large UK-based study pointing at an incidence of pulmonary hypertension as high as 7% in these patients.22

Although less studied to date, some reports have also suggested that neonates treated with diazoxide are at significantly increased risk of NEC. Of the seven studies found in our literature review, five had an increase in incidence of NEC compared with what would be expected for their included populations. While the exact pathophysiology is unknown, possible mechanisms that have been proposed for this increased risk include potassium channel-mediated hyperpolarisation of bowel smooth muscle cells and subsequent gut dysmotility,6 and reflex splanchnic vasoconstriction that would occur secondary to the systemic vasodilatory properties of diazoxide.8

It is unclear whether diazoxide dose may demonstrate a dose response in the development of NEC, since there is significant heterogeneity in the reported studies. The use of higher doses has been described as a possible modifying factor for the pathogenesis of pulmonary hypertension secondary to diazoxide.19 23 With regard to relationship of NEC with higher diazoxide dosing, two of the included studies that reported the highest incidence of NEC6 8 used median dosages of 10 mg/kg/day in their respective patient cohorts, and the only study that did not report any cases of NEC in a cohort of late preterm infants used doses of 3–5 mg/kg/day.7 This suggests that there might be a correlation of NEC with higher diazoxide dosage that needs to be evaluated in future prospective or large retrospective cohort studies. The duration of the treatment is quite variable among the studies; hence, it is difficult to draw conclusions from this: while the study by Chandran et al 7 reported a longer mean duration of 66 days, which is similar to previous studies,24 25 two other included studies reported much shorter courses,2 8 and the study by Grey et al 2 only counted the days of medication until discharge from hospital. Another interesting gap in knowledge is the timing and characteristics of presentation in these infants, with Prado et al 6 describing postmenstrual age of 36 weeks at presentation.

From a practical perspective, diazoxide use is still needed, given the increasing incidence of hyperinsulinism in preterm and term neonates and the lack of efficacious and safe alternatives for enteral treatment in these patients. Commencement of diazoxide should always be in consultation with paediatric endocrinology, and the decision to start or increase in dosing should be carefully discussed on a case-by-case basis, both with parents and the neonatal team, considering previous reports of diazoxide-induced pulmonary hypertension and concerning data regarding NEC in moderate and late preterm infants. Possible treatment alternatives especially in preterm and small for gestational age neonates include longer duration of high-concentration dextrose infusions, glucagon infusions,26 longer duration of feeds and carbohydrate modules (maltodextrin based powder) to increase enteral GIR.

There are several limitations associated with this review. The number of studies included is limited; most of them had a small study sample, and none of them had more than 10 cases of NEC; the level of evidence is low, given the retrospective nature of the included studies and the significant heterogeneity found among studies. Most of the included studies neither adjusted for other risk factors or other medications that could impact the risk of NEC nor included a control group of similar but non-exposed infants to better characterise the increase in risk. Naturally, more high-quality studies are needed to better understand the risk profile of diazoxide in neonates regarding NEC and other possible complications.