Clinical Advances in Pediatric Endocrinology: Focus on: Hyperinsulinism (long version)

Introduction Update on Diagnosis and Treatment of Hyperinsulinemic Hypoglycemia Mark A. Sperling, MD, and Charles A. Stanley, MD Case Analysis and Discussion Charles A. Stanley, MD, and Mark A. Sperling, MD

Introduction

Hyperinsulinemic hypoglycemia is the most dangerous form of hypoglycemia. Insulin affects not only glucose metabolism, but also the metabolism of protein and fat. Notably, insulin suppresses the conversion of fat to ketone bodies, which are the ultimate source of fuel for muscles and the brain under circumstances of fasting and low glucose. Thus, hyperinsulinism not only results in hypoglycemia, but also prevents the body from providing any alternative source of fuel to substitute for glucose. For this reason, hyperinsulinemic hypoglycemia carries a very high risk of causing irreversible brain damage unless promptly diagnosed and treated. Some forms of the disorder respond well to medical management, but others may require near-total pancreatectomy to control blood glucose levels.

The molecular mechanism and pathology of congenital hyperinsulinism has taken approximately 50 years to understand, with particularly exciting developments occurring over the past 10-15 years. For example, the molecular mechanisms by which the chemical energy of glucose is coupled to the activity of the ß-cell and the secretion of insulin have been significantly unraveled. This knowledge has subsequently led to insight on how the pathway of insulin secretion may become dysregulated. There is also the discovery that ß-cells in the pancreas take up L-3,4-dihydroxyphenylalanine (L-DOPA), and that this uptake is increased in areas of the pancreas that harbor potentially curable focal lesions. This discovery applied to newer imaging techniques has resulted in the ability to characterize and localize a lesion before surgery and offers the potential for complete cure while leaving most of the pancreas intact. Despite these advancements, however, there is still room for improvement. For example, only about 50% of cases of hyperinsulinism are accurately diagnosed and the frequency of developmental delay and seizures continues to be between 25% and 50%.

This monograph will serve to update pediatric endocrinologists and pediatric endocrine nurses who care for children who have hypoglycemia. The history of the field will be reviewed, along with the potential causes of hyperinsulinism. Diagnostic methods and treatment strategies will also be discussed. Furthermore, a case example will illustrate how to apply knowledge on diagnosis and treatment directly to practice.

I thank Dr. Charles A. Stanley for his article, case, and discussions, which led to the development of this monograph. Readers can expect to be provided with the most up-to-date knowledge of the field of hyperinsulinism, as well as strategies to use this knowledge to improve their practice.

Mark A. Sperling, MD
Course Chair

Course Chair: Mark A. Sperling, MD Professor and Chair Emeritus Department of Pediatrics Children’s Hospital of Pittsburgh Pittsburgh, Pennsylvania

Charles A. Stanley, MD Professor of Pediatrics University of Pennsylvania School of Medicine Director of the Hyperinsulinism Center Core Laboratory Director Clinical Translational Research Center The Children’s Hospital of Philadelphia Philadelphia, PA

Update on Diagnosis and Treatment of Hyperinsulinemic Hypoglycemia

Mark A. Sperling, MD, and Charles A. Stanley, MD

Introduction

Hypoglycemia due to genetic and non-genetic forms of hyperinsulinism (HI) make up the most common, as well as the most difficult to manage, disorders of hypoglycemia in pediatrics. Affected children are at high risk for seizures and irreparable brain damage if not promptly recognized and adequately treated. Some cases present as neonates with severe intractable hypoglycemia, but milder cases may escape detection until later in childhood or even until adult life. Recent discoveries have identified 8 genetic loci associated with congenital hyperinsulinism; these can have recessive, dominant, or sporadic patterns of inheritance. Testing for some of these genes is routinely available in commercial laboratories. In addition, there is increasing recognition that a prolonged neonatal form of hyperinsulinism can occur in association with disorders that cause perinatal stress, such as birth asphyxia and intrauterine growth retardation. Some forms of hyperinsulinism can be controlled well with medical therapy; others may require surgical near-total pancreatectomy. In more than half the cases that require surgery, the cause of the hyperinsulinism is an isolated focal lesion of the pancreas that is potentially curable by surgery. New methods for preoperative diagnosis and localization of such lesions, such as imaging by fluorine-18 L-3, 4-dihydroxyphenylalanine ([¹⁸F] L-DOPA) positron emission tomography (PET) scans, are becoming available.

Historical Note

Children with what we now recognize as congenital HI were first described in the early 1950s under the rubric of “idiopathic hypoglycemia of infancy.” In a subset of these cases, hypoglycemia could also be provoked by high-protein meals or by amino acids, particularly leucine. In the 1960s, the development of the first radioimmunoassays for insulin showed that idiopathic hypoglycemia was associated with dysregulation of insulin secretion. In the 1970s, it was suggested that these children had a problem in islet cell maturation evidenced by persistence of ß-cells budding off the ductal epithelium, which was termed nesidioblastosis. However, subsequent investigations showed that this pattern of nesidioblastosis is a normal feature of the pancreas during the first several months after birth so this term should no longer be used. By the 1990s, it was appreciated that congenital HI could be inherited in either recessive or dominant fashion and the first genetic loci associated with HI were described. As discussed below, these discoveries of the genetic basis of HI led to rapid advances in clinical diagnosis and management of this challenging group of disorders.^1-3

Clinical Features

Fasting hypoglycemia is the most important feature of HI. In severe cases, hypoglycemia occurs less than 1 or 2 hours after, or even in spite of, feeding. This reflects the potent action of insulin to increase the rates of glucose utilization while suppressing the rates of glucose production. Glucose infusion rates (GIR) needed to maintain normoglycemia in patients with HI usually are greater than 10–15 mg/kg/min and may exceed 20–30 mg/kg/min (5–6 times normal). Neonates with severe congenital HI are usually large-for-gestational age (LGA) at birth, due to the anabolic effect of insulin on fetal growth. Such infants are easily confused with infants born to diabetic mothers. Postprandial hypoglycemia is not usually a feature of HI; however, specific forms of HI are associated with protein-sensitive hypoglycemia, with or without leucine-sensitivity.^1-3

Diagnosis

There are several criteria for making the diagnosis of HI (Table). The best method of diagnosing HI is a provocative fasting test to monitor the fuel and hormone responses to hypoglycemia. Opportunity should also be taken at times of spontaneous hypoglycemia episodes to obtain specimens for diagnosis. In HI in children, the problem is dysregulated insulin secretion with a lack of adequate suppression of insulin, rather than oversecretion. Therefore, serum levels of insulin may not be clearly elevated at times of hypoglycemia, so emphasis must be placed on demonstrating markers of excessive insulin effects on circulating fuels (free fatty acid and ß-hydroxybutyrate) and the glycemic response to glucagon.^4,5 Point-of-care meters for measuring plasma ß-hydroxybutyrate, if available, in addition to blood glucose measurements by meter, are useful in evaluating suspected HI in real time. In some circumstances, additional biomarkers, such as insulin-like growth factor binding protein-1(IGFBP1), C-peptide, or proinsulin levels may provide useful information.⁶

Variants of Hyperinsulinism: Genetic Forms

There are many major genetic and non-genetic forms of HI. Since 1995, 8 genetic loci have been associated with congenital HI.^1,7 Defects in these loci involve steps in the pathways by which metabolic fuels “trigger” insulin release by increasing ATP levels to close the ß-cell ATP-dependent potassium (KATP) channels, leading to depolarization of the ß-cell plasma membrane (Figure 1). Note that diazoxide, the current drug of choice for medical treatment of HI, suppresses insulin secretion by binding to and activating KATP channel opening. Diazoxide is obviously not effective in patients with KATP channel mutations that cause complete loss of channel function. As discussed below, differences in phenotypes among the disorders can be helpful in pinpointing the underlying defect and defining therapeutic strategies.

Figure 1. Pathways of Insulin Secretion and Genetic Defects Causing Congenital Hyperinsulinism Glucose and amino acids stimulate insulin release by generating ATP, which leads to closure of ATP-sensitive plasma membrane potassium channels, plasma membrane depolarization, activation of voltage-sensitive calcium channels, an increase of cytosolic calcium, and release of insulin from storage granules. Source: Charles A. Stanley, MD

KATP-channel HI [inactivating mutations of the SUR1 (ABCC8) or Kir6.2 (KCNJ11) subunits of the ß-cell KATP channel; 11p15]

There are 2 subunits of the KATP channel: SUR1, which is encoded by the gene ABCC8, and Kir6.2, which is encoded by the gene KCNJ11. Inactivating mutations in the genes for either of these subunits result in inactive channels and consequently dysregulated insulin secretion. Mutations in the channel can be inherited in either a dominant or a recessive fashion.

Recessive KATP-HI

This is responsible for most of the cases of severe, diazoxide-unresponsive, diffuse HI. Infants are often born LGA and present with hypoglycemia at birth. GIR can be very high (20–30 mg/kg/min). Although some cases may be managed with octreotide plus intensive tube feeding regimens, many require surgical near-total pancreatectomy.

Focal KATP-HI

Congenital focal lesions of ß-cell adenomatosis are usually small, between 0.5 cm and 1 cm in diameter. They arise early in fetal development of the pancreas when a paternally-transmitted recessive KATP-channel HI mutation becomes duplicated due to loss of the normal maternal allele.⁸ A similar two-hit loss of heterozygosity mechanism is also responsible for many forms of cancer (eg, retinoblastoma). The phenotype is identical clinically to that of diffuse HI caused by recessive KATP mutations, ie, diazoxide is ineffective and affected infants often have high GIR and require surgery within a few weeks after birth.

Dominant KATP-HI

This is responsible for approximately half of the cases of diazoxide-unresponsive HI in whom a genetic defect can be found. Although some of those affected present with severe hypoglycemia in the neonatal period, other cases have few symptoms and may escape recognition even into adult life.⁹

GCK-HI (activating mutations of glucokinase; 7p15)

This is a relatively rare form of HI, but more cases continue to be reported, providing new information on the range of phenotype.¹⁰ The mutations are dominantly expressed, but approximately half of reported cases have de novo mutations. Although originally described as being responsive to diazoxide, it is increasingly recognized that plasma glucose may be difficult to normalize in this form of HI. This illustrates the fact that glucokinase (GCK) has a major effect on setting the ß-cell glucose threshold for insulin release.

GDH-HI (also known as the hyperinsulinism-hyperammonemia syndrome; caused by activating mutations of glutamate dehydrogenase, GLUD1; 10q23)

This is one of the more common diazoxide-responsive forms of HI for which mutations have been identified.¹¹ The mutations are dominantly expressed, but most patients have de novo mutations. Since glutamate dehydrogenase (GDH) is the site where leucine acts to stimulate insulin secretion, affected individuals are susceptible to protein-induced hypoglycemia and are hypersensitive to leucine-induced hypoglycemia. Patients have a high frequency of developmental delay and of absence seizures (generalized epilepsy), which seem not to be due to hypoglycemia, but may reflect effects of abnormal GDH function in the brain.^12,13

SCHAD-HI (inactivating mutations of short-chain 3-hydroxy-acyl-CoA dehydrogenase, HADH; 4q22)

This is a rare recessive form of diazoxide-responsive HI that is caused by a deficiency of a mitochondrial enzyme involved in ß-oxidation of short-chain fatty acids. The mechanism responsible for insulin dysregulation has not yet been elucidated. The disorder can be identified by metabolic tests showing increased plasma 3-hydroxy-butyryl-carnitine and urinary 2-hydroxy-glutarate.

In addition to the above, other rarer genetic HI defects have been described:

• MCT-1 (dominant exercise-induced HI; caused by mutations that interfere with silencing of ß-cell expression of a plasma membrane pyruvate carrier, monocarboxylate carrier 1, SLC16A1; 1p13.2).
• UCP-2 HI (dominant inactivating mutations of the mitochondrial ATP-carrier, uncoupling protein 2, UCP2; 11q13).
• HNF-4A (dominant inactivating mutations of the MODY1 gene, TCF14; 20q12); this defect later evolves into the MODY1 form of adult diabetes.⁷

Variants of HI: Acquired/Transient Forms

Perinatal stress HI

This is a transient form of congenital HI associated with birth asphyxia, small-for-gestational age (SGA) birthweight, or maternal toxemia/hypertension. Hypoglycemia may resolve within a few days but can last for several months. Most patients respond to diazoxide treatment. The mechanism is not understood, but the disorder is not uncommon and may occur in more than 10% of SGA infants.¹⁴

Other Types of Hyperinsulinemic Hypoglycemia or Mimickers of HI

Drug-induced HI and surreptitious insulin administration (Münchausen by proxy)

The possibility of hypoglycemia being induced by exogenous agents, such as oral sulfonylurea drugs (eg, glyburide) or injections of insulin, should be kept in mind even in very young children. Oral hypoglycemic agents may be ingested accidentally by young children, whereas insulin administration may be accidental or deliberate (syndrome of Münchausen by proxy). In the former setting, there will be release of both endogenous insulin and C-peptide secretion suggesting an insulinoma, whereas in the latter scenario, there will be suppressed levels of serum C-peptide at times of hypoglycemia [while serum insulin may be high or low, depending on the type of insulin administered (native human or animal insulin vs analogs of insulin) and their detectability with the assay employed]. Regardless, uncovering these causes requires the utmost vigilance and unusual perspicacity.

Insulinoma

Acquired tumors of the pancreas that release insulin are the most common cause of hypoglycemia in adults. Although rare, they occasionally occur in childhood. Most are benign, single islet cell adenomas; occasionally they may be associated with MEN1 mutations, which also can cause tumors of the pituitary.³

Congenital disorders of glycosylation (CDGs)

Some of the recessive disorders of glycosylation may have manifestations of hypoglycemia that involve or resemble HI. In some, such as CDG1a, the hypoglycemia responds to treatment with diazoxide; this particular defect interferes with synthesis of mannose and can be corrected with oral mannose supplementation. Others do not respond to mannose and also may not respond to diazoxide. The disorders can be screened for by testing for abnormal patterns of iso-electric focusing of plasma transferrin.^15,16

Autoimmune hypoglycemia disorders

Hypoglycemia resembling HI has been reported due to insulin receptor activating antibodies or with antibodies against insulin itself. Although rare, these disorders have occasionally been demonstrated in children as young as 1 year. Although autoimmune hypoglycemia disorders mimic the physiological features of HI, failure to detect elevation of serum insulin levels and failure to respond acutely to suppression of insulin release with octreotide may provide clues to these unusual disorders.¹⁷

Molecular Diagnosis

Commercial laboratory tests for the most common genetic forms of congenital HI are available (ABCC8, KCNJ11, GLUD1, and GCK). In addition, HNF-4A (TCF14) testing is available under tests for MODY-type diabetes. These tests provide direct sequencing of coding regions and portions of flanking introns and can be selected based on clinical phenotype (eg, for a diazoxide-unresponsive case, tests for ABCC8 and KCNJ11 will be useful, but GCK should also be considered). Testing is particularly useful for preoperative diagnosis of focal vs diffuse HI. Testing may be definitive for diffuse HI if two known disease-causing recessive mutations of ABCC8 or KCNJ11 are found. The demonstration of a paternal-only recessive ABCC8 or KCNJ11 mutation can suggest possible focal HI; however, it does not exclude possible diffuse disease.

Histology of focal vs diffuse HI

The histology of diffuse HI shows an abnormality consisting only of nuclear enlargement in 5% to 10% of ß-cells (3–4 times larger than surrounding cells).^18,19 This can be very subtle and requires a pathologist experienced in congenital HI to interpret correctly. The histology of focal HI shows a single area of ß-cell adenomatosis which may be as small as a few millimeters with normal-appearing islets in the surrounding tissue (Figure 2).

Figure 2. Pathology of Diffuse HI In diffuse HI, the islets throughout the pancreas are characterized by having enlarged nuclei in a few ß-cells. At low power (left panel), the islets from a diffuse HI case are normal in size and distribution. However, at high power (right panel), the islet in the center has one ß-cell containing a nucleus that is > 3 times larger than those of surrounding cells, which supports a diagnosis of diffuse HI, rather than focal HI. Source: Mariko Suchi, MD, PhD

Imaging for focal vs diffuse HI

Standard imaging procedures have not been useful in identifying or localizing focal adenomatosis lesions in infants (including CT, MRI, ultrasound, intra-operative ultrasound, octreoscan, etc). Invasive radiologic procedures are also not reliable, such as trans-hepatic portal vein insulin sampling or selective pancreatic arterial calcium stimulation/venous insulin sampling. Recently, good results have been obtained using [¹⁸F] L-DOPA PET scans.²⁰ While this remains an experimental procedure in most countries, efforts should be made to make it available prior to surgery in infants with congenital diazoxide-unresponsive HI, because 50% or more of cases will have potentially curable focal lesions.

Click here for more information on [18F] L-DOPA PET scan imaging.

Plasma Glucose Standards for Diagnosis and Treatment of HI and Other Hypoglycemia Disorders

The goal of treatment is to maintain plasma glucose levels above 70 mg/dL. Successful treatment adequate to allow for management at home means achieving this goal for periods of fasting of at least 8–12 hours. For provocative tests of hypoglycemia, a plasma glucose level of 50 mg/dL is low enough to identify abnormalities of fuel and hormone responses and unlikely to provoke serious symptoms of hypoglycemia (except in disorders of fatty acid oxidation). Note that the common practice of using lower standards for diagnosis and treatment of hypoglycemia in neonates vs older children and adults is strongly discouraged. This is especially true in HI, since the levels of ketones and other fuels are suppressed by insulin, leaving no alternative source of energy to support brain metabolism when plasma glucose is low.

Emergent Treatment of HI

At the finding of hypoglycemia, while treatment must be given immediately, consideration should also be given to obtaining essential diagnostic information just before starting treatment. This involves drawing an extra amount of blood, referred to as a critical sample, for later determination of important fuels (ketones and free fatty acid) and hormones (insulin, cortisol, growth hormone, etc). Intravenous dextrose, 200 mg/kg, can be given (2 cc/kg of 10% dextrose solution). Continuous infusion of dextrose should then be given at a level sufficient to maintain plasma glucose above 70 mg/dL.

Chronic Medical Treatment of HI

Diazoxide

Diazoxide is the drug of first choice for treatment of HI, since it acts directly to open the KATP channel and suppress insulin secretion. It is not effective in severe KATP channel inactivating defects and may not be completely effective in GCK-HI. Responsive patients require 5–15 mg/kg/d divided into 1 or 2 doses. Since the half-life is 24–36 hours, patients can only be judged as unresponsive after being on 15 mg/kg/d for 4–5 days. Effectiveness should be assessed by a 10–14 hour fasting test prior to discharge. Adverse effects include fluid retention (which may require diuretics) and hypertrichosis (excess hair growth in non-sexual areas, eg forehead and back).

Octreotide

Octreotide acts downstream of the KATP channel complex to suppress insulin release and may be useful for short-term or, occasionally, for chronic therapy in patients who fail to respond to diazoxide. Since octreotide downregulates its own receptor, tachphylaxis and desensitization to the effect of the drug on blood sugar control make octreotide effective only rarely for long-term use. It may be given as 3–15 µg/kg/d divided into 3–4 doses, preferably starting with low doses and titrating up as needed. Adverse effects include altered intestinal motility, steatorrhea, gall stones, and rare instances of necrotizing enterocolitis.

Other drugs

Calcium channel blockers, glucocorticoids, growth hormone, and other insulin antagonists have been used to treat infants with HI but are not effective and usually not worth trying. Glucocorticoids, especially, are not helpful in treating hypoglycemia due to HI.

Dietary treatment

Tube feedings (nasogastric or gastrostomy) may be useful in some cases as a continuous infusion to maintain normal plasma glucose levels. Force feeding should be avoided to prevent development of GERD and feeding refusal behaviors. Cornstarch is not helpful in management of HI. For some types of HI that are associated with protein-sensitive hypoglycemia, high protein feedings should be avoided.

Surgical Management of HI

Because of the need to distinguish focal from diffuse HI, surgery for congenital HI requires both surgical and pathology teams with experience in dealing with these cases.²¹ In most centers, the disorder is encountered only rarely; thus, cases should be referred to centers that have established expertise. Prior to surgery, results of mutation analysis and [¹⁸F] L-DOPA PET scans may provide an indication of whether the disease is focal or diffuse and, if focal, where the lesion is located. At surgery, biopsies for frozen sections are taken of 3 regions of the pancreas (head, body, and tail) to assess whether there is evidence of diffuse disease. If not, careful search by inspection and palpation is done to identify a focal lesion. For diffuse disease, a 98% near-total resection is necessary to assure that hypoglycemia control improves. For focal disease, the lesion is resected locally, leaving as much unaffected pancreas intact as possible. For lesions in the head of the pancreas, a Roux-en-Y procedure may be necessary.

References

1. De Leon DD, Stanley CA. Mechanisms of Disease: advances in diagnosis and treatment of hyperinsulinism in neonates. Nat Clin Pract Endocrinol Metab. 2007; 3:57-68.
2. Dekelbab BH, Sperling MA. Hypoglycemia in newborns and infants. Adv Pediatr. 2006; 53:5-22.
3. Kapoor RR,James C, Hussain K. Advances in the diagnosis and treatment of hyperinsulinemic hypoglycemia. Nat Clin Pract Endocrinol Metab. 2009; 5:101-112.
4. Finegold DN, Stanley CA, Baker L. Glycemic response to glucagon during fasting hypoglycemia: an aid in the diagnosis of hyperinsulinism. J Pediatr. 1980; 96:257-259.
5. Stanley CA, Baker L. Hyperinsulinism in infancy: diagnosis by demonstration of abnormal response to fasting hypoglycemia. Pediatrics. 1976; 57:702-711.
6. Levitt Katz LE, Satin-Smith MS, Collett-Solberg P, Thornton PS, Baker L, Stanley CA, Cohen P. Insulin-like growth factor binding protein-1 levels in the diagnosis of hypoglycemia caused by hyperinsulinism. J Pediatr. 1997; 131:193-199
7. James C,Kapoor RR,Ismail D,Hussain K. The genetic basis of congenital hyperinsulinism. J Med Genet. 2009; 46:289-299
8. de Lonlay P, Fournet JC, Rahier J, Gross-Morand MS, Poggi-Travert F, Foussier V, Bonnefont JP, Brusset MC, Brunelle F, Robert JJ, Nihoul-Fekete C, Saudubray JM, Junien C. Somatic deletion of the imprinted 11p15 region in sporadic persistent hyperinsulinemic hypoglycemia of infancy is specific of focal adenomatous hyperplasia and endorses partial pancreatectomy. Journal of Clinical Investigation. 1997; 100:802-807.
9. Pinney SE, MacMullen C, Becker S, et al. Clinical characteristics and biochemical mechanisms of congenital hyperinsulinism associated with dominant KATP channel mutations. J Clin Invest. 2008; 118:2877-2886.
10. Sayed S, Langdon DR, Odili S, et al. Extremes of clinical and enzymatic phenotypes in children with hyperinsulinism due to glucokinase activating mutations. Diabetes. 2009; 58:1419-1427.
11. MacMullen C, Fang J, Hsu BYL, et al. Hyperinsulinism / hyperammonemia syndrome in children with regulatory mutations in the inhibitory GTP binding domain of glutamate dehydrogenase. 2001; J Clin Endocrinol Metab. 86:1782-1787.
12. Kelly A, Stanley CA. Neurological aspects in hyperinsulinism-hyperammonaemia syndrome. Dev Med Child Neurol. 2008; 50:888.
13. Bahi-Buisson N, Roze E, Dionisi C, et al. Neurological aspects of hyperinsulinism-hyperammonaemia syndrome. Dev Med Child Neurol. 2008; 50:945-949.
14. Hoe FM, Thornton PS, Wanner LA, Steinkrauss L, Simmons RA, Stanley CA. Clinical features and insulin regulation in infants with a syndrome of prolonged neonatal hyperinsulinism. J Pediatr. 2006; 148:207-212.
15. Jaeken J, Matthijs G. Congenital disorders of glycosylation. Annu Rev Genomics Hum Genet. 2001; 2:129-151.
16. Jaeken J, Matthijs G. Congenital disorders of glycosylation: a rapidly expanding disease family. Annu Rev Genomics Hum Genet. 2007; 8:261-278.
17. Redmon JB, Nuttall FQ. Autoimmune hypoglycemia. Endocrinol Metab Clin North Am. 1999; 28:603-618.
18. Rahier J, Guiot Y, Sempoux C. Persistent hyperinsulinaemic hypoglycaemia of infancy: a heterogeneous syndrome unrelated to nesidioblastosis. Arch Dis Child Fet Neonatal Ed. 2000; 82:F108-F112.
19. Suchi M, Thornton PS, Adzick NS, et al. Congenital hyperinsulinism: intraoperative biopsy interpretation can direct the extent of pancreatectomy. Am J Surg Pathol. 2004; 28:1326-1335.
20. Hardy OT, Hernandez-Pampaloni M, Saffer JR, et al. Accuracy of [18F]Fluorodopa Positron Emission Tomography for Diagnosing and Localizing Focal Congenital Hyperinsulinism. J Clin Endocrinol Metab. 2007; 92:4706-4711.
21. Adzick NS, Thornton PS, Stanley CA, Kaye RD, Ruchelli E. A multidisciplinary approach to the focal form of congenital hyperinsulinism leads to successful treatment by partial pancreatectomy. J Pediatr Surg. 2004; 39:270-275.

Case Presentation

Charles A. Stanley, MD, and Mark A. Sperling, MD

Presenting Complaint

This 4 ½-month old girl was transferred from another hospital for treatment of hypoglycemia.

History

Birth

She was born to a G2P1, 28-year- old mother at term following an uncomplicated pregnancy, labor, and delivery. The birth weight was 7 lb, 2 oz.

On day of life (DOL) 3, the blood glucose was noted to be 20 mg/dL. The baby was observed to be hypothermic and lethargic, but otherwise did not appear to be in distress. The blood glucose rose to 45 mg/dL after an oral feeding of breast milk by bottle; subsequent blood glucose tests by bedside meter ranged from 40–70 mg/dL. The mother was advised that the low blood glucose was due to poor breast feeding. The baby was discharged from the nursery later that day without further investigations.

3 Months of Age: First Hospital

At 3 months of age, the baby was observed to have several staring spells and was taken to an emergency room, where the blood glucose was found to be 30 mg/dL by meter. The baby was admitted to the hospital and discharged after 2 days on diazoxide. At discharge, the mother was advised to feed the baby every 3 hours and check blood glucose levels at home with a meter.

Second Hospital

Over the next 2 weeks, the mother noted that the meter blood glucose was often in the 30 mg/dL range, although the baby appeared asymptomatic at these times. The baby was admitted to a second hospital for further diagnosis and treatment. Laboratory tests were obtained at this hospital (Table 1).

She was given a continuous infusion of dextrose by vein at a rate of 2 mg/kg/min and began treatment with subcutaneous injections of octreotide, 5 µg/kg/d. She was then referred to a third hospital for evaluation and treatment.

Third Hospital

On examination, the baby appeared healthy and developmentally appropriate for her age. She had no dysmorphic features and had no mid-line facial anomalies. Her length, weight, and head circumference plotted on the 50% percentile for age and were appropriate for the parental sizes. There were no unusual skin lesions. Examination of the heart and lungs was normal. The liver and spleen were not enlarged and no masses were palpated in the abdomen. The areolae and breast examinations were normal for her age. The external genitalia had a normal female configuration. The limbs were normal.

On admission to third hospital, the baby was receiving intravenous dextrose at a rate of 6 mg/kg/min, plus diazoxide and octreotide, and feedings every 3 hours. This maintained the blood glucose above 70 mg/dL. After the intravenous fluids were discontinued, the blood glucose fell to 36 mg/dL 3 hours after a feeding.

Laboratory values were obtained at that time (Table 2). Immediately after the blood tests were drawn, glucagon, 1 mg, was injected intravenously; blood glucoses were 42 mg/dL at 0 min, 75 mg/dL at +10 min, 85 mg/dL at +20 min, 95 mg/dL at + 30 min, and 137 mg/dL at +40 min. Additional tests showed normal Na, K, Cl, HCO₃, AST, ALT, BUN, triglycerides, uric acid, cholesterol, total carnitine, free carnitine, CBC, and urinalysis. A plasma acyl-carnitine profile by tandem mass spectrometry showed no abnormal elevations of fatty acyl-carnitine species.

A peripheral blood sample of DNA on the baby and each of the parents was sent for analysis of mutations in the coding regions of the following genes: ABCC8, KCNJ11, GCK, and GLUD1. A paternal ABCC8 mutation was found (potentially significant, but not previously identified to be disease-causing).

Diazoxide was discontinued. Blood glucose levels above 70 mg/dL were maintained on a combination of octreotide plus feedings every 3 hours. However, the baby could not maintain her blood glucose >70 mg/dL when fasted for 4–5 hours. An [¹⁸F] L-DOPA PET scan was performed which showed a small focal area of uptake in the inferior aspect of the distal pancreatic tail.

At surgery, biopsies from the head and body of the pancreas revealed no abnormalities. A small, firm mass was palpated in the tail which, upon resection of 10% of the pancreas, revealed an area of focal adenomatosis measuring 4 mm in largest diameter. After recovery from surgery, the baby was able to tolerate a fast of 16 hours before her blood glucose dropped to 52 mg/dL; at this time, the plasma ß-hydroxybutyrate had risen to 2.6 mM, the insulin was < 3 µU/mL, and the blood glucose rose to only 75 mg/dL 40 minutes after injection of glucagon 1 mg. The baby was discharged from the hospital 2 weeks following surgery.

Case Analysis and Discussion

Differential at Day of Life (DOL) 3

On DOL 3, the low blood sugar of this baby was assumed to be due to poor breast feeding. While a possibility, it is still important to be skeptical about this explanation and follow up with additional tests to ensure the blood sugar is maintained in a normal range. An important consideration is that symptoms associated with hypoglycemia can be misconstrued as less severe issues such as sleepiness, unwillingness to feed, or difficulty in arousal. Thus, a very high index of suspicion of hyperinsulinism (HI) is absolutely critical for physicians to make this diagnosis. If an infant has a blood sugar that is < 50 mg/dL, the clinician should immediately test for HI. Even though this baby did not manifest the hallmark features of HI hypoglycemia, such as large for gestational age, a blood glucose of 20 mg/dL should not be considered a one-time event and subsequently dismissed, especially when there are symptoms such as hypothermia, low temperature, and lethargy. It should not have taken 4 ½ months for this child to be diagnosed with HI and treated for hypoglycemia. When the baby was being fed adequately, it should have been confirmed that the blood sugar rose to approximately 70 mg/dL. Another test that could have led to an earlier diagnosis was to skip a feeding and make sure that the baby could maintain her blood sugar above 70 mg/dL for 6–8 hours. Discharging the baby without follow-up is one of the contributing factors to this delayed diagnosis.

Differential at 3 months of age

At 3 months of age, the baby was noted, again, to have symptoms of hypoglycemia, such as staring spells. In the emergency room, the blood sugar was 30 mg/dL and the baby was admitted to the hospital. HI was obviously suspected, as the baby was treated with diazoxide. The baby was discharged without follow-up to ensure that the diazoxide was effective. Consequently, the baby’s blood sugar was found by the mother to be 30 mg/dL, and the baby was taken to another hospital. Initiation of a treatment without confirming that the treatment is effective is regrettably a frequent occurrence. In contrast to rushing into treatment, the more appropriate action is to first procure what is referred to as the critical sample. The time when blood is taken to measure glucose levels is an opportunity to draw a few extra milliliters of blood and measure insulin along with other hormones and fuels. If the insulin level is high and ketones are suppressed, this supports the diagnosis of HI and the baby must be seen in an expert facility.

The differential diagnosis of a 3-month-old infant with persistent, recurring hypoglycemia is very extensive and includes a variety of hormone and metabolic disorders, as well as acquired problems that can cause hypoglycemia. However, with a critical blood sample, the differential diagnosis could be substantially reduced. One of the important causes of hypoglycemia in infancy is hypopituitarism.¹ This baby does not have an enlarged liver, so she is unlikely to have a glycogen storage disorder.² She has normal muscle strength, which excludes a severe fatty acid oxidation disorder.³ The fact that the baby appeared healthy and developmentally appropriate for her age is very reassuring.

Differential at Third Hospital

When the baby was admitted to the hospital she received intravenous dextrose at a rate of 6 mg/kg/min along with medication. The striking observation is that as soon as the intravenous fluids were discontinued, the blood glucose rapidly fell to 38 mg/dL within 3 hours after feeding. At 3 months of age, a baby should be able to fast for at least 8 to 12 hours after a feeding without becoming hypoglycemic. The timing of the drop in blood sugar levels was helpful in terms of ruling out some of the possibilities in the differential diagnosis. For example, in disorders of fatty acid oxidation, a patient does not become hypoglycemic until 10 or 12 hours after a feeding. Again, glycogen storage disorders have been ruled out, because the baby does not have a large liver. This baby appears to need high levels of exogenous glucose to maintain normal blood sugar levels, and increased glucose consumption is a hallmark sign of HI.

Interpretation of Laboratory Results at Third Hospital

The serum insulin level is undetectable (below assay limit), which demonstrates that a diagnosis of HI cannot be made based on insulin levels alone. Other markers, such as serum ketone levels, must be examined as well.

The bicarbonate level is normal, which indicates that the baby does not have a disorder that causes lactic acidemia or keto acidemia. Moreover, the absence of acidemia points toward HI, as insulin suppresses the ability to make keto acids from fatty acids when the blood sugar is low. This suspicion is corroborated by the remainder of the laboratory results (Table 2). The lactate level is normal at 1.1 mM. The ß-hydroxybutyrate is 0.1 mM, which is typical for the well-fed state but not the fasting state. In the fasting state, ß-hydroxybutyrate levels rise to between 2.5 mM and 3 mM; thus, it appears that inappropriate suppression of ketone production is occurring in this case. The fact that the C-peptide level was “normal” also suggests HI, because levels of C-peptide should be suppressed with other causes of hypoglycemia.

When a 1-mg dose of glucagon was injected intravenously, the baby’s blood glucose rose to 137 mg/dL within 40 minutes. This observation further supports the exclusion of a glycogen storage disorder, as glucagon causes the liver to convert stored glycogen into glucose. This brisk glycemic response to glucagon is a characteristic response for HI. The normal response for hypoglycemia is to utilize all of the glycogen in the liver. However, this cannot occur in hyperinsulinemic hypoglycemia, as insulin suppresses liver glycogen from being mobilized. The injection of glucagon overcomes this suppression. Thus, the observed glycemic response to glucagon further supports the possibility of HI.

The serum cortisol level of 6–7 µg/dL is not high enough to exclude the possibility of cortisol deficiency, so a pituitary or adrenal deficiency is still possible at this point. Moreover, it is important to keep in mind that these truly low cortisol levels in the presence of hypoglycemia do not exclude HI. Elevations of counter-regulatory hormone levels in response to hypoglycemia secondary to HI do not always occur. Thus, low or low-normal cortisol and growth hormone levels in this setting should not be interpreted definitively as abnormal.

A diagnosis cannot yet be made. The most likely explanation is HI, because such a robust glycemic response to glucagon would not be observed with cortisol deficiency. Overall, at this point, cortisol deficiency cannot be excluded, but the case is pointing very strongly toward HI. The results of a corticotropin-releasing hormone (CRH) test excluded cortisol deficiency (Table 2). Genetic analysis was used to confirm that this was in fact a case of HI.

Genetic Analysis

Determining which genetic form of HI is present in this case will point toward which type or types of treatments are likely to work. Mutational analysis takes several weeks to be completed. Therefore it is essential that this is performed as early as possible to optimize treatment for HI. The most important defects in the insulin secretion pathway are mutations of SUR1/Kir6.2, glutamate dehydrogenase, and glucokinase. These were the genes that were considered in the genetic analysis of this baby.

Mutation analysis is especially important in identifying patients who have potential focal KATP-HI lesions of the pancreas that may be cured by surgery. Genetic analysis of both the parents and the baby will define whether both parents are carriers of the KATP-HI mutation and whether the baby consequently has a chance of having a homozygous mutation. Genetic analysis of the parents will also help to determine if the baby is a compound heterozygote, in which case the father has a slightly different mutation than does the mother and the child has inherited a mutation from each of the parents. Homozygous and compound heterozygous KATP-HI mutations result in the most severe diffuse disease, where even a 95%–98% pancreatectomy may not be helpful. There is also an increasing number of KATP-HI mutations that are recognized as autosomal dominant, where the child needs to inherit only one allele from either of the parents in order to have HI. These types of mutations tend to result in less severe clinical manifestations and to be responsive to diazoxide treatment.

A paternal mutation in the ABCC8 gene, which encodes the sulfonylurea receptor (SUR1) subunit of the KATP channel, was found. Accordingly, diazoxide was discontinued, as this drug is ineffective when the channel is inactive. With a combination of octreotide plus feedings every 3 hours, levels of blood glucose >70 mg/dL were maintained; however, the baby could not maintain blood glucose levels > 70 mg/dL after fasting for 4–5 hours. These observations are not surprising, as the effects of octreotide tend to wane fairly rapidly and are only effective long-term in approximately 5%–10% of babies.

Having only one paternally derived mutation in the SUR1 KATP-channel gene makes it possible that the baby has a potentially curable focal lesion in the pancreas. This is a two-hit phenomenon that involves inheritance of an inactivating mutation from the father combined with a loss of the mother’s allele within the first few weeks of pregnancy. The result is a clone of cells in the pancreas that duplicates the father’s mutation. This always occurs with the father having the mutation, because the short arm of chromosome 11, which carries these KATP-channel genes, is imprinted with some genes on the mother’s chromosome that act as growth-suppressing genes. Loss of these growth-suppressing genes results in an overgrowth of a clone of cells that are isodisomic for the father’s potassium channel that is responsible for the HI. This phenomenon is more common than originally believed 10–15 years ago. These focal lesions account for 60% of babies with diazoxide-unresponsive HI.

The paternal mutation in SUR1 found in the baby is potentially significant but was not previously identified to be disease-causing. This finding is compatible with the possibility of a focal lesion, rather than diffuse disease, which would have been the case if both the father and mother were carriers of the mutation.

[¹⁸F] L-DOPA PET Scan

At this point, since medical therapy is ineffective and the mutational analysis suggests that a focal lesion may be possible, the next step is to perform an [¹⁸F] L-DOPA PET scan to accurately identify the location of the lesion. The development of this test over the last 10–15 years has been one of the most exciting advances in this field, as it immediately provides the surgeon with the exact knowledge of the area of the pancreas that must be removed.⁴

In this case, the PET scan revealed a small area of uptake in the anterior aspect of the distal pancreatic tail. A distal pancreatectomy in this region is fairly straightforward. Lesions that are in the head of the pancreas are much more complicated, as the surgeon must be able to do this while preserving the tail of the pancreas, which can sometimes require a Roux-en-Y procedure. During surgery, it must be confirmed that the patient has focal rather than diffuse disease. This required taking biopsies of the head and the tail of the pancreas during surgery and having them examined by a pathologist, who confirmed that these areas of the pancreas did not manifest any abnormalities. Ten percent of the pancreas, which contained an area of focal adenomatosis measuring 4 mm in largest diameter, was removed. These results further demonstrate the advances in this field, which make possible the detection and removal of a cell mass that measures less than one-half of a centimeter, while leaving 90% of the pancreas intact.

Post-Surgery Examination

After surgery, tests were performed to confirm that the surgery had cured the baby of HI. She was able to tolerate a fast of 16 hours before the blood glucose decreased to 52 mg/dL. At this time, the ß-hydroxybutyrate had risen to 2.6 mM, indicating that ketone production in the fasting state was not as suppressed as it had been prior to surgery. Insulin levels were < 3µU/mL. After 1 mg of glucagon was injected, the blood sugar rose to only 75 mg/dL. These results demonstrate that the baby has been cured of HI. No further endocrine follow-up is required, as there is no risk of hypoglycemia returning or the possibility of diabetes.

Conclusion

Despite all of the exciting developments in this field, approximately 50% of HI cases are not diagnosed and treated promptly. This case underscores the importance of making a diagnosis of HI as early as possible, to avoid the risk of irreversible brain damage. When a baby has low blood sugar, it is important for physicians to always have a high index of suspicion for HI. A critical sample should be obtained early in the process to measure insulin and other biomarkers, so that a diagnosis can be made and HI can be treated sooner rather than later. Moreover, it is important to send samples for genetic analysis early in the process, as it takes several weeks to get these results. With any HI treatment strategy, it is critical to ensure that it is effective by confirming that the baby can sustain a fast for 8–12 hours without developing hypoglycemia. All of these steps will result in the successful treatment of patients with HI.

References

1. Kelly A, Tang R, Becker S, Stanley CA. Poor specificity of low growth hormone and cortisol levels during fasting hypoglycemia for the diagnoses of growth hormone deficiency and adrenal insufficiency. Pediatrics. 2008; 122:e522-e528.
2. Ozen H. Glycogen storage diseases: new perspectives. World J. Gastroenterol. 2007; 13:2541-2553.
3. Rinaldo P. Fatty acid transport and mitochondrial oxidation disorders. Semin Liver Dis. 2001; 21:489-500.
4. Hardy OT, Hernandez-Pampaloni M, Saffer JR, et al. Accuracy of [¹⁸F]fluorodopa positron emission tomography for diagnosing and localizing focal congenital hyperinsulinism. J Clin Endocrinol Metab. 2007; 92:4706-4711.