Both Relative Insulin Resistance and Defective Islet β-Cell Processing of Proinsulin Are Responsible for Transient Hyperglycemia in Extremely Preterm Infants
Objective. Many extremely preterm infants develop hyperglycemia in the first week of life during continuous glucose infusion. The objective of this study was to determine whether defective insulin secretion or resistance to insulin was the primary factor involved in transient hyperglycemia of extremely preterm infants.
Methods. A prospective comparative study was conducted in appropriate-for-gestational-age preterm infants <30 weeks of gestational age with the aim specifically to evaluate the serum levels of proinsulin, insulin, and C-peptide secreted during transient hyperglycemia by specific immunoassays. Three groups of infants were investigated hyperglycemic (n = 15) and normoglycemic preterm neonates (n = 12) and normal, term neonates (n = 21). In addition, the changes in β-cell peptide levels were analyzed during and after intravenous insulin infusion in the hyperglycemic group. Data were analyzed using analysis of variance and analysis of variance for repeated measures.
Results. At inclusion, insulin and C-peptide levels did not differ in hyperglycemic subjects and in preterm controls. Proinsulin concentration was significantly higher in the hyperglycemic group (36.5 ± 3.9 vs 23.2 ± 0.9 pmol/L). Compared with term neonates, proinsulin and C-peptide levels were higher in normoglycemic preterm infants (23.2 ± 0.9 vs 18.9 ± 2.71 pmol/L and 1.67 ± 0.3 vs 0.62 ± 0.12 nmol/L, respectively). During and after insulin infusion in hyperglycemic neonates, plasma glucose concentration fell and proinsulin and C-peptide levels were lowered (18.4 ± 7.6 and 20.7 ± 4.5 pmol/L, respectively).
Conclusion. These data suggest that 1) preterm neonates are sensitive to changes in plasma glucose concentration, but proinsulin processing to insulin is partially defective in hyperglycemic preterm neonates; 2) hyperglycemic neonates are relatively resistant to insulin because higher insulin levels are needed to achieve euglycemia in this group compared with normoglycemic neonates. These results also show that insulin infusion is beneficial in extremely preterm infants with transient hyperglycemia.
Transient neonatal hyperglycemia is commonly observed during the first week of life in the extremely preterm infants (EPIs) <30 weeks of gestation. They develop hyperglycemia when receiving total parenteral nutrition despite a rate of glucose infusion that matches the basal requirement of the EPI (4–7 mg/kg per minute).1 Previous studies showed that plasma glucose and insulin concentrations were higher in preterm than in term neonates.2–4 However, in these studies, the radioimmunoassays used were unspecific: anti-insulin antibodies cross-reacted with proinsulin and intermediates. Only 1 study, using specific immunoradiometric assay, showed that concentrations of proinsulin and 32-33 split-proinsulin accounted for 34% to 70% of the total concentration of immunoreactive insulin and propeptides in euglycemic preterms.5 No study so far has assessed the various components of immunoreactive insulin in a group of hyperglycemic EPIs.
In the present study, immunoreactive proinsulin, insulin, and C-peptide secretion were specifically evaluated in appropriate-for-gestational-age EPIs. The aim was to determine whether defective insulin secretion or resistance to insulin was the primary factor involved in transient hyperglycemia, by comparing the secretory patterns of hyperglycemic EPIs, euglycemic EPIs, and normal term neonates. In addition, we analyzed the changes in proinsulin, insulin, and C-peptide concentrations during and after intravenous insulin infusion in the group of hyperglycemic preterm neonates.
The ethical committee of Cochin Hospital (Paris) approved this prospective and comparative study. Infants who were admitted before the sixth hour in the intensive care unit of the “Institut de Puériculture” from April 1999 through September 2000 were enrolled in this study when their gestational age was 24 to 30 weeks and their birth weight was appropriate for gestational age according to Mamelle et al.6 Informed written consent was obtained from the parents. Noninclusion criteria were sepsis, congenital anomalies, and intrauterine growth retardation. All included infants received total parenteral nutrition at an initial rate of 7 g/kg per day glucose (4.8 mg/kg per minute). The increasing daily rate was 1 to 2 g/kg per day. Refractory hypotension was treated with hydrocortisone (2 mg/kg first dose and 0.5 mg/kg/6 hours during 24–72 hours).7 Hyperglycemic subjects were enrolled when capillary glycemia tested by a reagent strip (Glucotrend; Roche, Meylan, France) was found at least twice above 11 mmol/L when tested every 1 to 2 hours. When hyperglycemia occurred, the glucose infusion rate was not lowered but hyperglycemic neonates received continuous intravenous insulin infusion (Insulin Actrapid; Novo Nordisk, Boulogne-Billancourt, France). The control subject for each hyperglycemic infant enrolled was the first preterm infant who was admitted in the unit with the same gestational age ±1 week and the same inclusion criteria except that plasma glucose concentration was between 4.3 and 7.6 mmol/L. A group of 21 term neonates who were sampled during the first week of life for assessing endocrine functions and found not to have any metabolic or endocrine disorder served as an additional control group.
Three blood samples were obtained in hyperglycemic subjects by nonperfuse artery or vein: 1) at enrollment, before insulin infusion; 2) during insulin infusion, when plasma glucose concentration was between 4.4 and 7.7 mmol/L for at least 8 hours; and 3) after the insulin was stopped and plasma glucose concentration was <10 mmol/L. One blood sample was obtained in control subjects at enrollment. Plasma glucose concentration was measured in each blood sample by the glucose oxidase method, using a quantitative glucose analyzer (Optima; Kone instruments, Espoo, Finland). For hormone assays, up to 1.5 mL of blood was taken on each occasion. Samples were centrifuged immediately and frozen at −20°C until assayed.
Serum insulin, proinsulin, and C-peptide of insulin were measured by means of immunoassays, which exhibited minimal cross-reactivities and required minimal assay volumes. Insulin was measured by a sandwich assay using Sanofi Bi-insulin immunoradiometric assay reagents (BioRad, Marnes la Coquette, France). The cross-reactivity of human proinsulin was <0.0001%, that of split 32-33 proinsulin and 31-32 proinsulin was <0.0004%, and that of C-peptide was <0.003%. Split 65-66 proinsulin and 64-65 proinsulin exhibited 100% cross-reactivity. The sensitivity was 0.2 μU/mL. Proinsulin was measured by radioimmunoassay using Linco Research reagents supplied by Nichols (Paris, France). The cross-reactivity of human insulin, human C-peptide, and 64-65 proinsulin was <0.1%. That of 31-32 proinsulin was 95%. The sensitivity was 4 pmol/L. C-peptide was measured by radioimmunoassay using Diagnostic System Laboratory reagents supplied by Brahms (St Ouen, France). The cross-reactivity of insulin was undetectable; that of proinsulin was 4%. The sensitivity was 0.0033 nmol/L.
Categorical data in 2 groups were compared by the χ2 test or Fischer exact test. For comparing continuous variables, 1-way analysis of variance or parametric Kruskall-Wallis test was used. Comparisons of insulin, proinsulin, and C-peptide in the hyperglycemic group over time were analyzed by analysis of variance for repeated measures and Bonferroni correction for multiple comparisons. Calculations were performed on the NCSS2000 software (Statistical Solutions, Kaysville, UT). Results are presented as mean ± standard error of the mean (SEM) unless otherwise stated.
Fifteen hyperglycemic (16 mmol/L [9.1–29 mmol/L]; mean [range]) and 12 control subjects (5 mmol/L [4.34–7.60 mmol/L]) were enrolled. At enrollment, both groups were similar in gestational age, birth weight, and distribution according to gender and respiratory disease. Use of exogenous surfactant, antenatal corticotherapy for pulmonary maturation, or postnatal corticosteroids for hypotension were also similar in both groups (Table 1). Maternal hypertension was noticed for 1 subject in each group; antihypertensive treatment was used only for the case in the control group. Gestational diabetes was noticed for 1 case in the control group. Glucose infusion rate (mean [range]) was higher in control than in hyperglycemic subjects (9.7 mg/kg per minute [6.9–15.5] vs 6.25 mg/kg per minute [4.2–12.5]; P = .00001) because median postnatal age (extremes) at enrollment was higher in control (5 days [3–8] vs 2 days [0–8]; P = .0005). Urinary glucose level was significantly higher in the hyperglycemic group (32.37 ± 5.13 mmol/L vs 6.58 ± 5.34 mmol/L; P = .002; Table 1).
In hyperglycemic subjects, insulin infusion was started with a mean dose of 0.22 mU/kg per minute (0.14–0.4 mU/kg per minute), and maximal mean dose was 1 mU/kg per minute (0.27–2.7 mU/kg per minute). The median duration of insulin infusion was 213 hours (37.5–656 hours). During insulin therapy, blood samples were collected in 9 infants between 12 and 98 hours after the treatment started (median: 37 hours). Last blood samples were collected between 59 and 645 hours after the insulin infusion was stopped (median: 224 hours), at a median postnatal age of 23 days (6–51 days).
At enrollment, immunoreactive insulin and C-peptide levels in hyperglycemic infants did not differ from those in control EPIs (13.9 ± 2.4 μU/L vs 11.6 ± 3.3 μU/L [P = .46]; 1.79 ± 0.3 nmol/L vs 1.67 ± 0.3 nmol/L [P = .78], respectively). Proinsulin levels were significantly higher in the hyperglycemic group than in control EPIs (36.5 ± 3.9 pmol/L vs 23.2 ± 0.9 pmol/L; P = 0006; Fig 1). In addition, immunoreactive insulin, C-peptide, and proinsulin levels were significantly lower in the control term neonates than in the control EPIs (P = .0009, P = .0003 and P = .023 respectively) and than in the hyperglycemic EPIs (P = .0012, P = .0009 and P = .0001, respectively; Table 2). In hyperglycemic infants, decreases in immunoreactive insulin levels over the enrollment (13.9 ± 2.4 μU/L) and the insulin therapy (10.4 ± 2.2 μU/L) and after insulin infusion was stopped (8.6 ± 1.4 μU/L) were not significant. Using analysis of variance for repeated measures, overall comparison of proinsulin mean levels between pretreatment, during treatment, and posttreatment fell significantly (36.5 ± 3.9 pmol/L vs 18.4 ± 7.6 pmol/L and 20.7 ± 4.5 pmol/L, respectively; P = .02). C-peptide level also varied significantly during the 3 periods (1.79 ± 0.3 nmol/L vs 0.57 ± 0.25 nmol/L and 1.2 ± 0.2 nmol/L, respectively; P = .002). A significant decrease in C-peptide was noted between enrollment and during the insulin therapy period (P < .01; Fig 2). Using analysis of variance for repeated measures, overall comparison of plasma glucose concentration between pre, during, and post insulin infusion fell significantly (16.56 ± 1.41 mmol/L vs 5.02 ± 0.35 mmol/L and 6.07 ± 0.49 mmol/L, respectively; P < .00001).
Small sample volume and specific immunoassays of various peptides secreted and processed by β-cell of the pancreas allow us to explore the reasons for hyperglycemia in EPIs. This study gives evidence that pancreatic β-cell secretion is impaired in hyperglycemic preterm neonates who are unable to secrete mature insulin in appropriate amounts and that hyperglycemic and control preterm neonates are relatively resistant to insulin.
Fifteen of 27 infants had glucose intolerance in the first week of life, although the rate of glucose administration matched the basal glucose requirement of the EPIs (4–7 mg/kg per minute).1 Higher urinary glucose concentration was noticed in the hyperglycemic EPIs as reported previously.8 At enrollment, insulin and C-peptide levels were similar in the hyperglycemic and in the control preterms, but proinsulin levels were significantly higher in the hyperglycemic group. In this last group, insulin infusion significantly decreased C-peptide and proinsulin levels. In addition, proinsulin, insulin, and C-peptide were significantly lower in term infants than in hyperglycemic and control preterm neonates.
In Hyperglycemic Preterm Infants, β-Cells Are Sensitive to Changes in Blood Glucose Concentrations
In these neonates, hyperglycemia induced very high levels of proinsulin (and possibly 31-32 proinsulin), although insulin levels did not differ significantly from those in control EPIs. After insulin infusion, proinsulin and C-peptide levels fell significantly in the hyperglycemic neonates, showing that the decrease in plasma glucose concentration was accompanied by a decrease in β-cell secretion. Altogether, these data give evidence that the mechanisms that govern the production of proinsulin-related peptides by the pancreatic β-cells are effective in preterm neonates.
In Hyperglycemic Preterm Infants, the Processing of Proinsulin to Insulin Is Defective
In response to hyperglycemia, β-cells of preterm infants only increased the secretion of the nonprocessed hormone proinsulin, which is at least 10-fold less active than insulin and does not allow controlling serum glucose level.9 This can likely be related to defective islet β-cell processing of proinsulin. That proinsulin levels in the control preterm group are also markedly higher than in the term neonates suggests that all preterm infants exhibit immature insulin processing at variable degree. This defective islet β-cell processing of proinsulin probably persists late in gestation as suggested by the high levels of proinsulin observed in this study when insulin therapy was stopped at a median postnatal age of 23 days. This is in agreement with the results of Hawdon et al,5 who found a large proportion of insulin intermediates during the first week of life in normoglycemic neonates between 25 and 34 weeks of gestation. Proinsulin is normally processed into mature insulin in the secretory vesicles of pancreatic β-cells that contain the required processing enzymes proconvertases PC2 and PC3. The proconvertases are neuroendocrine-specific proteases that are required for the processing of different hormones, such as the proopiomelanocortin, the precursor of corticotropin-releasing hormone. The deficient endocrine-processing system could explain some of the relative hormonal deficiencies that are reported in the preterm neonates.10
Hyperglycemic EPIs Are Partially Resistant to Insulin
Immunoreactive insulin was higher in hyperglycemic and control preterms than in term neonates. The total insulin immunoreactivity in the EPIs represents the sum of “true” insulin levels together with some proinsulin-conversion intermediates because the antibody system used cross-reacts with split 65-66 proinsulin and 64-65 proinsulin. Unfortunately, excessive blood subtraction would have been needed to allow chromatographic separation of these insulin precursors. Anyway, C-peptide level was significantly higher in both groups of EPIs compared with term neonates (Table 2). This clearly indicates that insulin production is higher in EPIs than in older infants and suggests that preterm infants are relatively resistant to insulin. In addition, hyperglycemic EPIs are more resistant to insulin than the control EPIs because similar amounts of insulin failed to achieve euglycemia. Although some of the EPIs received hydrocortisone (0.5 mg/kg/6 hours), the dosage was estimated in physiologic range as cortisol basal normal secretion is assumed to be 6.6 to 8 mg/m2 per day and stress production rate 4 times basal secretion (ie, 0.63–0.77 and 2.5–3.1 mg/kg per day, respectively).11,12 We believe that this treatment cannot be related to insulin resistance.
Many studies have shown that continuous insulin infusion improves glucose tolerance in extremely low birth weight infants, facilitates provision of calories, and enhances weight gain in glucose-intolerant premature infants.13–15 There are controversies in the use of insulin therapy in preterm infants because the reason for impaired glucose tolerance in extremely immature neonates still remains unclear.16,17 The relative insulin resistance observed in the population of extremely preterm infants reported here supports the concept that it would be beneficial to offer insulin substitutive treatment to hyperglycemic preterm neonates. Insulin levels should be increased by insulin infusion supra the biological levels to achieve euglycemia. The insulin dosage needed for controlling hyperglycemia in our patients was similar to those reported in the literature.18
Hyperglycemia May Result From Persistent Hepatic Glucose Production Related to Partial Insulin Resistance
By contrast with adults, hepatic glucose production persisted during parenteral glucose infusion in human preterm infants.19–21 The glucose transporter GLUT-2 allows glucose uptake or release by the liver in response to variation in plasma glucose concentration.22 Once in the hepatocyte, glucose is converted into glucose-6 phosphate by glucokinase, before entry in the various metabolic pathways. Glucokinase insulin-dependent activity may be decreased in the liver of the preterm neonates secondary to the relative defect in insulin activity and/or sensitivity.23 This should lead to excessive output of glucose in the circulation because hepatocytes are unable to metabolize glucose efficiently.
In addition, diminished peripheral glucose uptake by insulin-sensitive tissues such as adipose tissue and skeletal and cardiac muscle may contribute to hyperglycemia in the neonate. Unless not statistically different, hyperglycemic EPIs in this study had lower birth weight than euglycemic EPIs and probably less abundant insulin-sensitive tissues.
In conclusion, our study demonstrates that both defective processing of proinsulin in β-cells and partial resistance to insulin are present in hyperglycemic EPIs. These defects are likely responsible for altered glucose homeostasis. These data are in favor of continuous insulin infusion as a treatment of transient hyperglycemia in EPIs.
This work was supported in part by the Association pour le Développement de l’Hygiène Maternelle et Infantile (Paris) and by the Institut de Recherche Endocrinienne et Métabolique (Paris).
We are indebted to Yvette Le Bihan, Josiane Le Fourn for skillful technical assistance. We thank Jean Girard for fruitful discussion and advice. We gratefully acknowledge the help from the medical and nursing staff of the Institut de Puériculture (Paris).
- Received January 24, 2003.
- Accepted June 3, 2003.
- Reprint requests to (D.M.-M.) Service de Réanimation Pédiatrique et Néonatale, Hôpital Necker-Enfants Malades, 149 Rue de Sèvres, 75015 Paris, France. E-mail:
- ↵Hawdon JM, Aynsley-Green A, Alberti KG, Ward Platt MP. The role of pancreatic insulin secretion in neonatal glucoregulation. I. Healthy term and preterm infants. Arch Dis Child.1993;68 :274– 279
- ↵Pollak A, Cowett RM, Schwartz R, Oh W. Glucose disposal in low birth weight infants during steady state hyperglycemia: effects of exogenous insulin administration. Pediatrics.1978;61 :546– 549
- ↵Helbock HJ, Insoft RM, Conte FA. Glucocorticoid-responsive hypotension in extremely low birth weight newborns. Pediatrics.1993;92 :715– 717
- ↵Wilkins BH. Renal function in sick very low birth weight infants: 4. Glucose excretion. Arch Dis Child.1992;67 :1162– 1165
- ↵Revers RR, Henry R, Schmeiser L, et al. The effects of biosynthetic human proinsulin on carbohydrate metabolism. Diabetes.1984;33 :762– 770
- ↵Kanarek KS, Santeiro ML, Malone JI. Continuous infusion of insulin in hyperglycemic low-birth weight infants receiving parenteral nutrition with and without lipid emulsion. J Parenteral Enteral Nutr.1991;15 :417– 420
- ↵Iynedjian PB, Jotterand D, Nouspikel T, Asfari M, Pilot PR. Transcriptional induction of glucokinase gene by insulin in cultured liver cells and its repression by the glucagon-cAMP system. J Biol Chem.1989;264 :21824– 1829
- Copyright © 2004 by the American Academy of Pediatrics