Objective. The aim of this study was to investigate sequential neuroradiologic changes in the brains of infants after transient neonatal hypoglycemia. We used magnetic resonance imaging (MRI) and ultrasonography (US) head scans.
Methods. Eighteen symptomatic full-term infants whose serum glucose concentrations were ≤45 mg/dL (2.5 mmol/L) without any other diseases were included in the hypoglycemic group. MRI and US head scans were performed at full-term age and at the age of 2 months. The imaging results were compared with the findings of MRI and US scans in 19 healthy normoglycemic term newborn infants at the respective ages. The neurologic outcome was followed in the both groups.
Results. MRI or US showed evidence of abnormality in 39% the hypoglycemic infants. MRI detected more abnormalities in the brains than US. Four infants showed patchy hyperintensity lesions either in the occipital periventicular white matter or the thalamus on T1-weighted images. These lesions had a good tendency to recover and only 1 of these infants appeared to be neurologically affected. Of the 19 controls, 10% (2 of 19) had caudothalamic cysts, which were detected both with MRI and US. The relative risk of the hypoglycemic child compared with nonhypoglycemic child, to have any abnormality detected in the brain, was 3.7, with a 90% confidence interval from 1.11 to 12.28.
Conclusions. Postnatal full-term MRI and US scans showed abnormalities four times more often after transient neonatal hypoglycemia than in the healthy control group. However, most often lesions were absent 2 months later. The clinical relevance of these abnormal findings remains to be clarified with detailed neurologic examinations and follow-up.
- MRI =
- magnetic resonance imaging •
- US =
- ultrasonography •
- SGA =
- small for gestational age •
- RR =
- relative risk •
- CI =
- confidence interval
Glucose, like oxygen, is essential for the normal brain to function. Neonatal hypoglycemia occurs mostly when the normal processes of metabolic adaptation after birth fail to occur. It is a common metabolic and endocrine abnormality in growth-retarded infants and in infants of diabetic mothers. Severe neurologic sequelae have been reported after symptomatic neonatal hypoglycemia.1 The brain damage caused by hypoglycemia is documented histopathologically,2 in both adults3 and infants.4 ,5 Reports on the findings with computed tomography or magnetic resonance imaging (MRI) in infants with classic transient hypoglycemia, however, are sparse and consist mostly of patients with seizure activity after severe prolonged hypoglycemia.6–8
In the present study, the frequency and the distribution of cerebral MRI and ultrasonography (US) findings after neonatal hypoglycemia were recorded. The sensitivity of the two modalities was compared and the short-term prognosis of the imaging findings controlled at the age of 2 months with reference to respective clinical findings as a short-term report of a prospective clinical and imaging study of high-risk infants.
SUBJECTS AND METHODS
All the infants with neonatal hypoglycemia, admitted to the neonatal intensive care unit of the University Hospital of Turku, were imaged. Between February and October 1996, 18 neonates with symptomatic hypoglycemia, were enrolled in a prospective cranial US and MRI study (Table 1). Of the 18 infants, 6 were small for gestational age (SGA) and 2 were infants of diabetic mothers. The infants were born at 36 to 42 gestational weeks (Table 1). The definition of neonatal hypoglycemia included low serum glucose concentration (≤45 mg/dL or 2.5 mmol/L), associated with clinical manifestations like tremor, apathy, tachypnea, irritability, hypotonia, and/or difficulties in feeding. The symptoms disappeared after therapy that restored the blood glucose to normoglycemia. The infants with only one low serum glucose value before the age of 6 hours were excluded. The patients were also placed into one of two categories, according to the recurrence of neonatal hypoglycemia, ie, (I) two or more hypoglycemic episodes, and (II) one hypoglycemic episode after the age of 6 hours. The infants were free from congenital malformations and they had no infections. There was a control group of 19 healthy neonates, born at term (Table 1). Pregnancy and delivery were uncomplicated for these infants.
Hypoglycemia was treated with oral feedings and intravenous glucose. Oral feedings were started at 1 to 3 hours of age with 10- to 20-mL volumes, and continued at 3-hour intervals. The volume was increased by 10 to 20 mL each day until the total daily volume reached one-fifth of body weight. After the diagnosis of hypoglycemia, continuous infusion of glucose was administered at the rate of 6 to 7 mg/kg/min as 10% glucose. If levels of plasma glucose could not be maintained, the rate of glucose administration was progressively increased to 10 to 15 mg/kg/min. Plasma glucose concentrations were measured regularly at 1- to 3-hour intervals to determine the efficacy of the therapy. If symptoms recurred or persisted or if serum glucose concentration of >45 mg/dL (2.5 mmol/L) could not be maintained after 4 to 6 hours of 10 to 15 mg/kg/min, hydrocortisone (5 mg/kg/day every 12 hours) was added to the regimen. Once stable, plasma glucose was monitored at 4- to 6-hour intervals preprandial. Healthy control infants were breastfed normally and had no glucose supplementation or medication.
Neonatal medical information and demographic data taken from patients' charts included infant gestational age, birth weight, length, Apgar scores at 1 and 5 minutes, and umbilical arterial pH. The cranial MRI and US examinations were performed both at term and at the corrected gestational age of 2 months. The developmental assessment included a careful neurologic examination by a pediatrician and a physiotherapist, at 2, 4, 6, 8, and 12 months' corrected ages. The neurologic and developmental evaluation based on the status of gross and fine motor functions, speech development, social behavior, sensory screening, and emergence of autonomy and independence. The study has been approved by the Ethics Committee of the Turku University Central Hospital. Parental consent was obtained after oral and written information.
Magnetic Resonance Imaging
MRI scans were performed on open 0.23-T MRI equipment (Outlook, Picker Nordstar, Helsinki, Finland). A multipurpose flexible coil fitting the head of the infant was used. Axial T2-weighted multislice spin-echo images with a repetition time (TR) of 3300 msec and echo time (TE) of 200 msec, a flip angle of 90°, a slice thickness of 7 mm, and a field of view of 139 × 220 mm were used to access the brains of the infants if the infant was <2 months of age. In infants >2 months of age, TE was 170 msec, but the remaining parameters were the same as mentioned before. A coronal T1-weighted field echo sequence with TR of 35 msec, TE of 12 msec, flip angle of 40°, slice thickness of 5 mm, and field of view of 220 × 220 mm were obtained also. The images were estimated by a neuroradiologist without information of the clinical status. The infants were imaged during postprandial sleep without any anesthesia. The open MRI equipment allowed visual control and easy access to the infant, and the infants were monitored by using a peripheral oximeter during the imaging. The findings were read blindly a second time by the same neuroradiologist (R.P.), to reveal intraobserver variability.
All patients went through initial US examination with a 7-MHz vector transducer (Acuson [Mountain View, CA] 128 XP/10C), during the perinatal period on the neonatal intensive care unit. Follow-up examinations were performed at 2 months age with a 5-MHz vector transducer (Aloka [Aloka Co Ltd, Tokyo, Japan] SSD 2000), in the Radiography Department. The examinations were performed by a pediatric radiologist through the anterior fontanelle. They were recorded on a videotape (SUPERVHS) and analyzed blindly by the same pediatric radiologist (H.R.), to reveal intraobserver variability.
Serum glucose concentrations were measured by an enzymatic method, using glucose-6-phosphate dehydrogenase with a detection limit of 1.8 mg/dL (0.1 mmol/L). Date are expressed as mean ± SD values. Demographic data were analyzed with the two-sample Studentt test. Differences were considered significant atP < .05. For abnormal findings, the relative risks (RRs) and their 95% confidence intervals (CIs) were calculated to quantify the significant associations.9 Fisher's two-tailed exact test was used. The statistical computation was performed with an SAS statistical program package.10
The detailed demographic data of the patients and healthy infants are presented in Table 1. The serum glucose concentration of the hypoglycemic infants was 25 ± 12 mg/dL (1.4 ± 0.7 mmol/L) and 5 infants needed hydrocortisone in addition to parenteral glucose infusion to cure the hypoglycemia. In the healthy asymptomatic infants, the serum glucose concentrations were 63 ± 13 mg/dL (3.5 ± 0.7 mmol/L) at the age of 66 ± 29 hours.
US and MRI scans
The hypoglycemic infants were examined with US and MRI both in the neonatal period (age, 37–42 postconceptional weeks) and 2 months later (age, 48–51 postconceptional weeks). There were 2 exceptions, ie, 1 patient had the second MRI and US examinations already at the age of 40 gestational weeks. In another patient the second MRI and US were done at the age of 6 months. The first MRI and US were performed in all 18 hypoglycemic infants. The second MRI was performed in 78% and the second US in 94% of these infants. The first MRI and US examinations were performed within 24 hours of each other in 72%, and the second examinations similarly in 67%, of hypoglycemic infants.
Cerebral MRI and US scans of the healthy infants were performed in the neonatal period (38–44 postconceptional weeks) and also 2 months later (47–49 postconceptional weeks) (Table 1). The first MRI and US examinations were performed at the neonatal period to all healthy infants. The follow-up MRI scans were performed in 63% and the second US scans in 95% of these healthy infants. The first MRI and US examinations were performed within 24 hours of each other in 95% of infants, and the second examinations, respectively, in 63% of infants.
Seven of 18 hypoglycemic infants (39%) had abnormal MRI and/or US examinations either at the neonatal period or at the age of 2 months. The abnormal cerebral imaging findings of the hypoglycemic children are shown in Table 2. At the term age, 33% (6 of 18) of infants had an abnormal MRI scan. Four infants had patchy hyperintense lesions either in occipital periventricular white matter or in the thalamus on T1-weighted images (Fig 1). One of these 4 infants also had hyperechogenic areas in the periventricular white matter, interpreted as leukomalasia at term age. In addition, 2 infants had unilateral dilatation of lateral ventricles. The RR of the hypoglycemic child compared with nonhypoglycemic child, of having any abnormality detected in the brain, was not statistically significant at the 5% level but is significant at the 10% level (RR = 3.7; P = .062; 90% CI, 1.11–12.28). These lesions had good tendency for recovering. To find possible confounding factors in occurrence of abnormalities, we studied the association of several prognostic factors within the hypoglycemic group (infant of diabetic mother, SGA, prematurity, and recurrent hypoglycemia). Occurrence of the abnormalities was quite similar in different classes of prognostic factors and these factors cannot account for the observed risk.
The distribution of the low serum glucose concentrations were compared with the MRI and the US imaging findings. Forty-two percent (5 of 12) of infants with two or more episodes of low serum glucose concentrations (≤45 mg/dL or 2.5 mmol/L) had abnormalities in the MRI or in the US findings and they did not differ from the newborns with only one hypoglycemic episode, 33% (2 of 6) (95% CI, 0.33–4.65). No relation was found between the SGA infants and the imaging findings; 33% (2 of 6) of the hypoglycemic SGA children had abnormalities in either the MRI or the US imaging findings. The RR of abnormal cerebral findings in hypoglycemic SGA infants was 0.8 (95% CI, 0.21–2.98).
Of the 19 controls, 10% (2 of 9) had caudothalamic cysts, which were detected with both MRI and US. One of these infants had cysts only at the age of 39 postconceptional weeks. The other infant had cysts at the age of both 39 postconceptional weeks and 2 months; these cysts were minimal in size.
One healthy infant had slight prominence of the occipital horn of the left ventricle at term age, on both MRI and US. In addition 2 infants had slight prominence of the left lateral ventricle at the age of 2 months on US (MRI examination could not be performed on these infants at the age of 2 months because of the restlessness of the children). However, the slight asymmetric prominence of the lateral ventricles in these 3 infants was considered as insignificant and their lateral ventricle indexes were within normal limits.
The development of the hypoglycemic children was followed and the mean follow-up time was 11 months (range, 5–12 months until now, and the follow-up will continue), and 94% (17 of 18) of infants had normal development (gross and fine motor functions, speech development, social behavior, sensory screening, and emergence of autonomy and independence). So far only 1 of these infants appeared to be neurologically affected. He had developed right-sided hemiplegia, and both of his MRI examinations showed abnormalities (Tables 1 and 2, patient 13). He had tremors, but after the treatment with parenteral glucose his symptoms disappeared. His first cerebral US scan at the age of 6 days was normal.
In the control group, the neurologic and physical examination appeared to be within normal limits during the neonatal period as well as during the follow-up. However, these healthy infants have as yet had a shorter follow-up period, until 4 months of age.
The occipital predominance of MRI abnormalities was similar to those previous studies on transient neonatal hypoglycemia.6–8 In our patients, hypoglycemia was rapidly and properly treated. This may have been the reason that we did not see such serious MRI findings as were described in earlier reports.6–8
The pathologic findings of the severe prolonged neonatal hypoglycemia are documented,4 ,5 but the imaging results are few. Spar et al6 presented the computed tomographic and MRI findings of 1 infant at the age of 19 days, after 15 hours of severe hypoglycemia. They found progressive parenchymal loss and predominantly occipital involvement. Barkovich et al8 reported, from computed tomography and MRI, similar patterns of injury observed in 5 neonates as a result of severe prolonged hypoglycemia with seizure activity. In addition, an MRI report on a hypoglycemic infant at the ages of 10 days and 4 months, from a letter by Aslan and Dinc,7 showed diffuse parenchymal loss and hypointense areas resembling infarction in both occipital regions and dilatation of the occipital horns of the lateral ventricles. At the age of 4 months there was atrophy of the occipital cortex. The authors postulated that selective occipital vulnerability may be related to intense axonal migration and synaptogenesis, which occurs within the occipital lobes during the neonatal period. Moreover, the likelihood of ischemia to certain brain regions in newborn dogs during hypoglycemia relates to the impairment of vascular autoregulation. Those brain regions included the cerebral cortex, hippocampus, and thalamus, each of which exhibits particular vulnerability to hypoglycemic neuronal injury.11
In a previous work we studied the effect of neonatal hypoglycemia on the local cerebral metabolic rate for glucose. We used positron emission tomography and 2-[18F]fluoro-2-deoxy-D-glucose as a tracer. The cerebral metabolic rate for glucose in patients after hypoglycemia was similar to the age-adjusted control infants and they had a normal neurodevelopmental outcome during the follow-up. We could not find any regional involvement.12 However, hypoglycemia was less severe than in the fatal cases, in which there were neuropathologic findings.4 ,5
Hypoxic/ischemic conditions in preterm and term infants are known to cause brain lesions that can be detected on MRI images as hyperintensities on T1-weighted images, and pathologically they represent reaction of glial cells and macrophages to hypoxia.13 In our study, 4 infants showed patchy hyperintensity lesions either in the occipital periventicular white matter or in the thalamus, on T1-weighted images.
According to our study, hypoglycemia can cause lesions similar to those caused by hypoxemia in the periventricular white matter or thalamus in term infants. These lesions have a good tendency to recover. Only 1 infant with the most severe lesions had lesions left at the age of 6 months (Table 1 and 2, patient 13). That infant had only slight and short hypoglycemia. Neurologic follow-up study demonstrated hemiplegia, which probably was not caused by hypoglycemia, but possibly by a prenatal vascular insult. This is suggested by posthemorrhagic subependymal hyperintensities in the left centrum semiovale on repeat MRI at the age of 6 months.
Two of 19 healthy infants showed caudothalamic cysts at term, and in 1 of these they persisted to the age of 2 months, but the cysts were smaller. Caudothalamic cysts are reported to appear as sequelae of germinal matrix hemorrhage,14 but they are known to be found in 5% of normal neonates15 and this finding is assumed to result from normal lysis of subependymal germinal matrix14 and is not a pathologic sign.
In our study, MRI detected more abnormalities in the brains of hypoglycemic infants than US. The superiority of MRI over US in detecting nonhemorrhagic parenchymal brain injury is also reported in other studies.16 Sonography has low sensitivity to cortical injuries. Therefore, parietooccipital injuries consequent to neonatal hypoglycemia are difficult to detect with US.
Up to 8% of low-risk infants suffer an episode of hypoglycemia, typically at 3 to 4 hours after delivery.17 The incidence of central nervous system damage in symptomatic patients has been reported to be as high as 50% to 60%.1 ,18 However, our patients do not compare with those diagnosed and treated 10 to 30 years ago. The recovery in our cases with hypoglycemia has been good during the follow-up so far. Their onset of hypoglycemia was relatively early, the duration was brief, and the degree of hypoglycemia was quite mild. Most of our hypoglycemic patients had symptomatic transitive-adaptive hypoglycemia and there were also some with classic transient neonatal hypoglycemia.19 Previous studies indicate that the neonatal brain is capable of using a variety of organic metabolites, for example, lactate,20 ketone bodies, and glycerol,21 as a replacement for energy production during hypoglycemia. Regardless of these protective mechanisms, prompt recognition and treatment of symptomatic neonatal hypoglycemia is necessary to minimize sequelae.
These children will be followed to obtain data of long-term neurodevelopmental outcome. It is unlikely that these infants will develop cerebral palsy, but minor neurologic problems will be sought during the planned follow-up of these infants for at least 6 years.
This study was supported by the South-West Finnish Fund of Neonatal Research.
We thank the participating families. We also thank the personnel at the Departments of Pediatrics and Diagnostic Radiology; Mrs Kati Saarinen, physiotherapist, for her assistance in the neurological follow-up of the patients; and Mrs Satu Ekblad, RN, for practical assistance.
- Received June 1, 1998.
- Accepted September 22, 1998.
Reprint requests to (A.K.) Department of Pediatrics, University of Turku, FIN-20520 Turku, Finland.
- Anderson JM,
- Milner RDG,
- Strich SJ
- Spar JA,
- Lewine JD,
- Orrison WW
- Barkovich JA,
- Al Ali F, Rowley HA, Bass N
- ↵Altman DG, ed. Practical Statistics for Medical Research. London, England: Chapman and Hall; 1997
- ↵SAS Institute Inc. SAS/STAT User's Guide, Version 6, Vols 1 and 2. 4th ed. Cary, NC: SAS Institute; 1990
- Shen E-Y,
- Huang F-Y
- Haworth JC,
- McRae KN
- ↵Cornblath M, Schwartz R, eds. Disorders of Carbohydrate Metabolism in Infancy. 3rd ed. Cambridge, UK: Blackwell Scientific Publications; 1991
- ↵Volpe JJ. Hypoglycemia and brain injury. In: Volpe JJ, ed. Neurology of the Newborn. 3rd ed. Philadelphia: WB Saunders; 1995:467–489
- Copyright © 1999 American Academy of Pediatrics