Functional Hyperactivity of Hepatic Glutamate Dehydrogenase as a Cause of the Hyperinsulinism/Hyperammonemia Syndrome: Effect of Treatment
Objective. The combination of persistent hyperammonemia and hypoketotic hypoglycemia in infancy presents a diagnostic challenge. Investigation of the possible causes and regulators of the ammonia and glucose disposal may result in a true diagnosis and predict an optimum treatment.
Patient. Since the neonatal period, a white girl had been treated for hyperammonemia and postprandial hypoglycemia with intermittent hyperinsulinism. Her blood level of ammonia varied from 100 to 300 μmol/L and was independent of the protein intake.
Methods. Enzymes of the urea cycle as well as glutamine synthetase and glutamate dehydrogenase (GDH) were assayed in liver tissue and/or lymphocytes.
Results. The activity of hepatic GDH was 874 nmol/(min·mg protein) (controls: 472–938). Half-maximum inhibition by guanosine triphosphate was reached at a concentration of 3.9 μmol/L (mean control values: .32). The ratio of plasma glutamine/blood ammonia was unusually low. Oral supplements withN-carbamylglutamate resulted in a moderate decrease of the blood level of ammonia. The hyperinsulinism was successfully treated with diazoxide.
Conclusion. A continuous conversion of glutamate to 2-oxoglutarate causes a depletion of glutamate needed for the synthesis of N-acetylglutamate, the catalyst of the urea synthesis starting with ammonia. In addition, the shortage of glutamate may lead to an insufficient formation of glutamine by glutamine synthetase. As GDH stimulates the release of insulin, the concomitant hyperinsulinism can be explained. This disorder should be considered in every patient with postprandial hypoglycemia and diet-independent hyperammonemia.
Neonatal or infantile hypoketotic hypoglycemia with persistent hyperammonemia has been associated with only a few inherited conditions, ie, either defects of the carnitine transport system such as carnitine acylcarnitine carrier deficiency1 or unusual forms of hyperinsulinism.2 In affected patients coma and/or convulsions may rapidly develop. Therefore, immediate adequate treatment is required.
The more common cases of isolated hyperinsulinism may be the result of genetic defects in the insulin secretion by the pancreatic β cells.3 These defects may reside in the plasma membrane sulfonylurea receptor (SUR 1) or an associated potassium channel (Kir 6.2) of the pancreatic β cells. Other causes of hyperinsulinism have also been described. These classic types of hyperinsulinism are not accompanied by elevated blood ammonia levels.
Recently, 10 patients with hyperinsulinism combined with hyperammonemia (HIHA) have been described.2,4–10 Both sporadic and dominantly inherited cases were observed. The clinical histories more or less resembled those of leucine-sensitive hypoglycemia,11 although disorders related with the ammonia disposal have not been reported in the latter condition. The HIHA syndrome has been studied in lymphoblasts. This syndrome was shown to be either the result of hyperactivity of glutamate dehydrogenase (GDH)10 or impaired inhibition of GDH by guanosine triphosphate (GTP).5 GDH catalyzes the conversion of glutamic acid into 2-oxoglutaric acid.
In this article, we report our attempts to treat a patient with HIHA in whom the primary defect could be established in liver tissue.
The female patient was born as the second child of healthy, nonconsanguineous parents after an uneventful pregnancy and delivery. Neither parent had ever experienced protein intolerance or a tendency to hypoglycemia. Her birth weight was 3840 g (75th centile [P75]), head circumference 36 cm (P75), and Apgar score 8/9. She was admitted on the twelfth day of life because of convulsions and cyanosis, associated with hypoglycemia. During the hypoglycemic episodes, which recurred after discontinuing the parenteral glucose supply, no ketone bodies were found in the urine. Hyperinsulinism was diagnosed on the finding of insulin levels of 28 and 45 μU/L at glucose concentrations of 2.8 and 2.7 mmol/L, respectively. When measured at low blood glucose levels, the 24-hour C-peptide excretion was 8.1 nmol (controls: 1.0–5.5). Blood ammonia was repeatedly elevated and varied from 100 to 200 μmol/L in the initial period (neonatal controls <80). All other routine clinical chemical analyses were normal.
An electroencephalogram did not show any epileptic changes, a computed tomography scan was performed at the age of 2 months and revealed widened ventricles as well as hypodensities of the frontotemporal white matter.
Feeding difficulties became a major clinical problem, necessitating gavage feeding from the age of 6 weeks. She was discharged at 14 weeks. At that time, treatment consisted of diazoxide (2 × 8 mg/day), a high-carbohydrate, and a protein-restricted diet [1.5 g/(kg·day)]. Arginine (and later citrulline, 5 × 1000 mg/day) and sodium benzoate (4 × 1500 mg/day) were administered to reduce the blood ammonia levels.
At the age of 2 years, the patient was referred to the University Children's Hospital and her condition was reevaluated. Her length was 84.5 cm (50th centile [P50]), body weight was 15 kg (>P90), and head circumference was 48 cm (P50). Her motor development was 7 months delayed; she showed a marked retardation of speech development. Blood ammonia was 312 μmol/L (controls: <50); no further clues to a diagnosis were found. Despite the high ammonia levels no clinical signs of hyperammonemia were noted (such as drowsiness, lethargy or ataxia). A brief trial with an extreme dietary protein restriction [.5 g/(kg·day)] did not have any effect on the ammonia levels. Plasma-free carnitine was 12 μmol/L (controls: 25–60), total carnitine 13 μmol/L (controls: 30–65). Carnitine supplements (2 × 750 mg/day) were administered. The treatment was adjusted by replacing the oligoglucose formula by uncooked cornstarch. Over the years, the diazoxide treatment was gradually adjusted to the requirements, reaching a dose of 4 × 50 mg/day at the age of 6 years.
As a result of the mitigated carbohydrate intake and the resulting decrease of fat deposits, the patient's adipose appearance markedly improved. However, her development showed a slow progress and the feeding difficulties persisted. These problems led to the decision to install a gastrostoma at age 5. During this intervention, a surgical liver biopsy was taken for diagnostic purposes. After the final diagnosis, treatment with oral supplements of carbamylglutamate (CG) (4 × 500 mg/day) was started. Blood ammonia levels tended to be lower on this regimen, but did not normalize completely. At the age of 6.5 years, her psychomotor development was estimated to be at a level of 4.5 to 5 years (not formally tested).
MATERIALS AND METHODS
Quantitative amino acid analysis was performed with a Biochrom 20 amino acid analyzer (Amersham Pharmacia Biotech, Cambridge, United Kingdom). Organic acids in urine were analyzed by gas chromatography/mass spectrometry (MD800, Fisons, Manchester, United Kingdom) of the ethoximated methyl esters after ethylacetate extraction. Orotic acid in urine was measured using a colorimetric assay.12 Profiling of plasma acylcarnitines was performed by fast-atom bombardment mass spectrometry after solid-phase extraction and butyl ester formation.13 All other clinical chemical analyses, including all aspects of insulin homeostasis, were performed by standard methods.
Glutamine synthetase activity was assayed in lymphocytes, leukocytes, and liver homogenate of the patient according to Rowe14with slight modifications. The synthesis of 1-14C-glutamine from 10 mM 1-14C-potassium glutamate was measured. The reaction mixture contained 10 mM phosphoenolpyruvate, 50 μM imidazole-hydrochloric acid (pH 7.2), 10 mM 2-mercaptoethanol, 1 mM ethylenediaminetetraacetic acid (pH 7.2), 20 mM NH4Cl, 20 mM MgCl2, 10 mM sodium adenosine triphosphate (pH 7.4), 5 μg (1 IU) pyruvate kinase and 3 μM rotenone. The total volume was 100 μL; incubation at 37°C was conducted for 30 or 60 minutes. After stopping the reaction with 400 μL of ice cold 4 μM imidazole-HCl (pH 7.2) plus 10 μM glutamine, the reaction product was isolated on a .5 × 7 cm AG-1x8 (200–400 mesh) ion exchange column, which was in the acetate form. GDH activity was assayed in a liver biopsy homogenate, essentially as described by Wreszcynski and Colman.15 The reaction of 2-oxoglutarate with ammoniumchloride forming glutamate and the simultaneous oxidation of NADH was followed spectrophotometrically at 340 nm. Stimulation of the enzyme activity was achieved by adding 200 μM adenosine diphosphate (ADP), whereas the effects of the regulatory GTP were studied at concentrations of .01 to 10.0 μM.
Ornithine carbamoyltransferase and carbamoyl phosphate synthetase activities were measured in a liver biopsy specimen by spectrophotometric methods.
Because of the persistent hyperammonemia, detailed studies of the urea cycle were performed. Both fasting and postprandial plasma and urine amino acid concentrations were always normal. In particular, citrulline, arginine, and ornithine were not decreased, However, the ratio of plasma glutamine versus blood ammonia was unusually low (Fig 1), unlike that observed in patients with various urea cycle defects.16 Plasma glutamate averaged 54 ± 25 μmol/L (controls: 30 ± 14), and was therefore not decreased. Urinary orotic acid excretions were inconsistent; at no occasion did we find a striking increase. Urinary organic acids, including 2-oxoglutarate, have always been normal. Cerebrospinal fluid (CSF) amino acids revealed a normal glutamine level (421, 732, and 800 μmol/L on 3 occasions, controls: 614 ± 241) and markedly increased levels of ammonia (100, 149 and 161 μmol/L, controls <11). The patient was treated with various regimens aimed at reducing the ammonia level. Neither a drastic reduction of the dietary protein to .5 g/(kg·day), nor supplementation with sodium benzoate and arginine or citrulline, led to changes of blood ammonia levels (Fig 2). In addition, treatment with neomycin remained without effect (data not shown).
At various occasions, simultaneous measurements of blood glucose and insulin were performed, clearly indicating hyperinsulinism. Early morning blood glucose was always in excess of 3 mmol/L; glucose levels during the day went down to 2.7 mmol/L with a concomitant insulin of 45 μU/L. On one occasion, blood glucose and insulin were checked before and after a meal. Glucose was 2.8 mmol/L before and 3.5 mmol/L 2 hours after the meal; simultaneous insulins were 28 and 101 μU/L, respectively.
The activities of hepatic carbamoyl phosphate synthetase and ornithine carbamoyl transferase were 463 nmol/(h·mg protein) (control: 587) and 36.9 μmol/(h·mg protein) (control: 45.5), respectively; both values were normal. Glutamine synthetase was assayed in leukocytes and liver, which resulted in values of .175 nmol/(min·mg protein) and 7.15 nmol/(min·mg wet weight) respectively. Control values for leukocytes and liver were .115 and 3.41, respectively.
The results of the GDH assay in liver are shown in Fig 3. Under standard conditions with ADP stimulation, the patient had a normal activity [874 nmol/(min·mg protein), controls 472–938 (n = 3)]. After addition of the inhibitor GTP, this activity decreased more slowly than in the control. Half-maximum inhibition was reached at a GTP concentration of 3.9 μmol/L (control mean: .32). Histologically, the liver tissue was normal; there were minimal signs of fibrosis. GDH in fibroblasts was normal [19.4 nmol/(min·mg protein); controls 7.9–26.8 (n = 3)]. The GTP inhibition was not tested in these cells.
Following the diagnosis, the patient was given a trial with oral CG (4 × 500 mg/day). This treatment was based on the hypothesis that she might suffer from a shortage of glutamate necessary for the synthesis of N-acetylglutamate. Blood ammonia levels were regularly checked during 10 months of follow-up and were found to be consistently lower than before the onset of the treatment (Fig 2).
Recurrent periods of hypoglycemia in infancy without the concurrent availability of alternative energy substrates such as ketone bodies or lactate, represents one of the most important metabolic abnormalities in young children. The same applies for hyperammonemia. A relatively mild increase of blood ammonia levels for short periods of time presumably leads to cerebral damage. Our patient, having only a moderate mental retardation, has had hyperammonemia with ammonia levels in blood and CSF of 100 to 300 μmol/L for 6 years in combination with episodically occurring nonfasting hypoketotic hypoglycemia. The latter was in fact the first presenting symptom, accompanied by convulsions. Subsequently hyperinsulinism was diagnosed and this was treated successfully with diazoxide. It was clearly shown that the hypoglycemias were most prominent after a protein-containing meal, similar to the observations made in leucine-sensitive hypoglycemia.11
As there were no diagnostic clues to the hyperammonemia, treatment was originally directed at alleviating the symptoms by administrating citrulline (originally arginine) for stimulation of the urea cycle, and sodium benzoate for the direct removal of ammonia in the form of hippurate. When we found the aberration in the GDH, implying that in this patient the function of the urea cycle was not compromised, conventional treatment was stopped, and replaced by CG treatment.
Recently, the HIHA syndrome has been identified.2 Not long thereafter, Stanley et al5 disclosed regulatory mutations of the GDH gene in some of the patients. Thus far, 3 out of 10 patients, whose clinical history has been published, showed the GDH insensitivity toward GTP in lymphoblasts. The patient reported by Yorifuji et al10 had a different mutation resulting in a permanent hyperactivity of GDH. No studies in liver were reported. Our data originating from liver tissue show that the basal GDH activity in the patient was normal, in accordance with the lymphoblast findings in sporadic cases.5 In contrast, the basal GDH-activity in familial cases was moderately decreased. The half-maximum inhibitory concentration of GTP in the patient's liver was approximately 10-fold higher than in control liver tissue. This difference was more pronounced than previously found in the patients' lymphoblasts.5 Presumably, GTP is present in hepatic tissue at a level that is sufficient to inhibit normal GDH. Mutant GDH—needing a much higher concentration of GTP for its inactivation—is supposed to be fully active when the GTP level is not decreased. There are only a few conditions in which the intracellular GTP concentrations are decreased, eg, hypoxanthine phosphoribosyltransferase deficiency (Lesch-Nyhan syndrome) or purine nucleoside phosphorylase deficiency. Neither of these conditions is associated with a deranged GDH activity leading to hyperinsulinism or hyperammonemia.
Hepatic glutamine synthetase activity in our patient's liver was increased. This may be considered as a compensatory mechanism for the removal of ammonia. However, glutamine synthetase is a high-affinity, low-capacity enzyme and is apparently not capable of removing all excess ammonia. The CSF ammonia levels in our patient were quite high, although accompanied by normal glutamine levels. This result again raises the question of ammonia/glutamine toxicity to the brain. We have the impression that a single elevation of brain ammonia—without glutamine—showed a less deleterious effect on the brain than a combined increase. Hyperammonemic rats, treated with the glutamine synthetase inhibitor methionine sulfoximine, had less brain edema than untreated hyperammonemic rats.17 Other reports have shown that brain glutamine synthetase is a saturable enzyme and that ammonia encephalopathy progresses only after saturation of this enzyme.18 It remains to be established whether brain glutamine synthetase in the present patient was actually exhausted. We do not have information about CSF ammonia levels in patients with urea cycle defects.
The clinical spectrum of HIHA, as deduced from our patient and the literature,2,4–10 is rather uniform. None of the patients was severely retarded. In general, moderate mental retardation or even normal development was reported. Convulsions—possibly the result of hypoglycemia—were invariably the presenting symptom. Although symptoms are expected to start in the first week of life, most patients were detected between 2 and 7 months. Blood ammonia levels were comparable in all patients, ranging from 75 to 350 μmol/L. None of the authors reported abnormalities of plasma amino acid profiles. There were 2 reports on increased 2-oxoglutarate levels in the urine.8,9 It was not mentioned whether these patients had a decreased sensitivity of GDH inhibition by GTP. So there seems to be some biochemical heterogeneity of the HIHA syndrome. Thus far, all patients with a proven defect of GDH had a normal 2-oxoglutarate excretion, including the patient with a permanent hyperactivity of GDH reported by Yorifuji et al.10
Treatment of the hyperinsulinism varied from simple diazoxide administration (as in the present patient) to subtotal pancreatectomy. It has been shown that glutamate stimulates insulin secretion in rats.19 The mechanism of this reaction is not entirely clear, but an enhanced oxidation of glutamate by GDH may be a contributing factor.5 Recently, it has been shown that insulin secretion can be stimulated by leucine, which activates GDH.20
The differential diagnosis of infantile nonketotic hypoglycemia has been extended. It is now clear that in every case of suspected hyperinsulinism blood ammonia levels should be determined, preferably together with plasma amino acids. The HIHA syndrome may prove to be less rare than originally believed.
We thank Dr O. P. van Diggelen (Erasmus University Rotterdam) for the assay of the urea cycle enzymes; N. van Rossum-de Jong and L.M. Hussaarts-Odijk for the assay of GDH and glutamine synthetase; and R. Jankie for the amino acid analyses.
- Received November 10, 1999.
- Accepted January 19, 2000.
Reprint requests to (J.G.M.H.) Erasmus University, Department of Clinical Genetics, Box 1738, 3000 DR, Rotterdam, The Netherlands. E-mail:
- HIHA =
- hyperinsulinism/hyperammonemia •
- GDH =
- glutamate dehydrogenase •
- GTP =
- guanosine triphosphate •
- CG =
- carbamylglutamate •
- ADP =
- adenosine diphosphate •
- CSF =
- cerebrospinal fluid
- Permutt MA,
- Nestorowicz A,
- Glaser B
- Rogers LE,
- Porter FS
- Bertrand G,
- Puech R,
- Loubatieres-Mariani MM,
- Bockaert J
- Gao ZY,
- Li G,
- Najafi H,
- Wolf BA,
- Matschinsky FM
- Copyright © 2000 American Academy of Pediatrics