Safe use of the ketogenic diet in an infant with microcephaly, epilepsy, and diabetes syndrome: a case report

Microcephaly, epilepsy, and diabetes syndrome (MEDS) is also known as microcephaly with simplified gyral pattern, epilepsy, and permanent neonatal diabetes syndrome (OMIMID 614,231) [1, 2]. It is associated with congenital microcephaly with progressive head growth deceleration, a simplified gyral pattern on brain MRI, infantile onset of epilepsy including infantile spasms, refractory neonatal diabetes mellitus, and severe global developmental delays. MEDS is an autosomal recessive condition caused by pathogenic variants in the IER3IP1 gene on chromosome 18q21 which encodes Immediate Early Response 3 Interacting Protein 1 (IER3IP1), a protein found in developing brain cortex and beta cells [2]. Bi-allelic loss of function variants in this gene lead to apoptosis in neurons and beta cells resulting in microcephaly, epilepsy, and insulinopenic diabetes [2]. MEDS has previously been described in 8 cases from Argentina, Morocco, and Egypt (Table 1) [1‐4]. We present a proband with bi-allelic IER3IP1 variants in trans and phenotypic features consistent with MEDS including microcephaly, insulinopenic diabetes mellitus, and a developmental and epileptic encephalopathy. Additionally, we report the use of both steroids and the ketogenic diet (KD) to ameliorate epilepsy symptoms, which presented an additional challenge in managing their insulin-dependent diabetes.

The proband was born via normal spontaneous vaginal delivery at 37 weeks gestations after an uncomplicated prenatal history as well as pregnancy and delivery. She is the second-born to a non-consanguineous couple and her older brother is healthy and typically developing. At birth, growth parameters were normal (weight 2.715 kg, height 50 cm) except head size at 8th percentile (Fig. 1). Her newborn screening tests and development was normal for the first 2 months when she was taken to the emergency department for new onset seizures and found to have frequent multiple focal seizures on electroencephalogram (EEG). Magnetic resonance imaging revealed no parenchymal signal abnormalities but did demonstrate prominence of subarachnoid spaces and small bilateral subdural hygromas along the temporal convexities. Her seizures evolved to include infantile spasms as well as focal seizures and EEG was notable for diffuse encephalopathy, multiple focal epileptiform discharges, and hypsarrhythmia at the time of spasm onset – consistent with an epileptic encephalopathy. A multigene epilepsy panel was sent as part of her diagnostic work-up (Supplemental Table) reported two variants of uncertain significance in IER3IP1 (NM_016097.4). The first, c.239T > G (p.Leu80*), is novel and predicted to result in loss of function through protein truncation or haploinsufficiency due to nonsense-mediated decay. The second, c.2T > A (initiator codon), likely impacts transcription as it affects the start codon of the IER3IP1 gene. Both variants meet the American College of Medical Genetics criteria for pathogenicity [5]. Sequencing the unaffected parents, obtained per clinical recommendations of the genetics team caring for the proband, demonstrated paternal inheritance for c.239T > G(p.Leu80*) and maternal inheritance for c.2T > A.

At 4 months of age, she was readmitted for infantile spasms and started high dose steroids. During this time, her blood glucose increased from 92 to 112 mg/dL before admission to 241-298 mg/dL at discharge. While on steroids, her parents reported an improvement in her general movements, eye tracking, and frequency of spasms. One week after discharge, steroids were discontinued due to persistent hyperglycemia (> 400 mg/dL) and despite multiple anti-seizure medications, seizures frequency, duration, and intensity increased. Thus, she was readmitted at 4.5 months of age to restart steroids and insulin initiation as her blood glucose levels increased to 200-400 mg/dL, consistent with beta cell dysfunction seen in MEDS. Blood glucose levels normalized on 0.35–0.7 units/kg/d of insulin treatment. Her blood glucose values were monitored with an intermittently scanned continuous glucose monitoring (isCGM).

We present a case of an infant with MEDS due to two bi-allelic pathogenic variants in IER3IP1 and refractory epileptic encephalopathy managed with steroids and the KD. Her management was complicated by a delicate balance between achieving ketosis for therapeutic purposes but avoiding severe ketosis which can result in DKA, cerebral edema, and other unwanted complications. In her clinical management, the difficulty is that the therapeutic effect of ketones for seizure prevention is fundamental in the pathway for causing DKA [7].

For ketone production (either therapeutic for the sake of seizure control or pathologic in the setting of DKA) there are three major mechanisms. First, decreasing carbohydrate intake and increasing fat intake results in fats being the primary fuel source (i.e., KD). Second, inducing a state of starvation requires utilization of metabolic stores (i.e., fat). Third, insulinopenic states result in an inability to inhibit lipolysis resulting in the production of ketones. Regardless of the mechanism of ketone production, ketosis with significant acidosis can lead to significant clinical complications including cerebral edema. At the time of KD initiation, our patient had a very low insulin requirement of only 0.1 units/kg/day. Given this low insulin requirement paired with the requisite limits to carbohydrates of the KD, she was considered to be at risk for euglycemic DKA. Furthermore, initiation and maintenance of therapeutic KD requires the production of ketones, and therefore, some level of initial acidosis.

One of the most challenging aspects of initiating the KD in the setting insulinopenic diabetes in our case was identifying safe biochemical ranges. While we use standard bicarbonate cutoff for DKA to set bicarbonate goals, the goal hydroxybutyrate level was much less clear because of the patient’s unique physiology and pathophysiology. For this reason, we opted to start with targets informed by prior case reports. Ultimately, we found that the patient was safe when her capillary ketones were ≤6mmol/L, higher than our initial anticipated goal for an infant with insulinopenic diabetes but the standard range for KD. We also demonstrated that despite the use of insulin, which prevents lipolysis and ketogenesis, steady state ketosis can be achieved in an infant on KD. To ensure safety, we found that isCGM/CGM and twice daily capillary ketone monitoring provided adequate triggers to escalate care.