| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Other Original Article |
Department of Intensive Care Medicine (D.M., F.W., G.V.d.B.) and Laboratory for Experimental Medicine and Endocrinology (F.V.), University Hospital Gasthuisberg, Catholic University Leuven, B-3000 Leuven, Belgium; and Kolling Institute of Medical Research, University of Sydney, Royal North Shore Hospital (D.M., P.J.D.D., K.V.H., R.C.B.), St. Leonards, New South Wales 2065, Australia
Address all correspondence and requests for reprints to: Greet Van den Berghe, M.D., Ph.D., Department of Intensive Care Medicine, University Hospital Gasthuisberg, B-3000 Leuven, Belgium. E-mail: greta.vandenberghe{at}med.kuleuven.ac.be.
Abstract
IGF binding protein-1 (IGFBP-1), an important regulator of IGF bioavailability, has been shown to correlate with mortality in critically ill patients. In the liver, IGFBP-1 is transcriptionally repressed by insulin, and it is therefore a potential marker of hepatic insulin sensitivity. We have recently shown that, compared with conventional treatment, maintenance of normoglycemia with intensive insulin therapy decreased morbidity and mortality of continuously fed critically ill patients. This study compares the effect of conventional and intensive insulin therapy on IGFBP-1 and assesses its predictive value for mortality.
In 363 patients who were dependent on intensive care for more than 7 d and were randomly assigned to either conventional or intensive insulin therapy, serum IGFBP-1 levels were measured on admission, on d 1, 8, 15, 22, and 29, and on the day of intensive care unit discharge or death. In addition, IGFBP-1 and phosphoenolpyruvate carboxykinase mRNA levels were measured by real-time RT-PCR in postmortem liver biopsies obtained from 74 patients who died in the intensive care unit.
Although intensive insulin treatment lowered glycemia, it had no effect on IGFBP-1 serum levels. Instead, serum IGFBP-1 concentration was significantly higher in patients who ultimately died, and it differentiated nonsurvivors from survivors 3 wk before death. The predictive value of serum IGFBP-1 for mortality was similar to that of the APACHE-II score. Like circulating IGFBP-1, hepatic mRNA levels of IGFBP-1 and the similarly insulin-regulated gene, phosphoenolpyruvate carboxykinase, were not significantly different between conventional and intensive insulin therapy groups.
These data suggest that hepatic insulin resistance in prolonged critically ill patients, reflected by high serum IGFBP-1 levels, is not overcome by intensive insulin therapy, and that this may affect patient outcome.
STRESS OF CRITICAL illness evokes insulin resistance and hyperglycemia. In a recent large study involving 1548 intensive care unit (ICU) patients, it was shown that maintaining normoglycemia (
6.1 mmol/liter or 110 mg/dl) during critical illness, using intensive insulin treatment combined with feeding, reduced ICU mortality by 43% (1). Furthermore, morbidity was reduced, and the need for prolonged mechanical ventilation and ICU dependency was to a large extent prevented.
Preliminary data suggested that a high serum concentration of IGF binding protein (IGFBP)-1 in the chronic phase of critical illness predicts ICU death (2, 3, 4). The IGFBP-1 gene, principally expressed in hepatocytes, is strongly regulated at the transcriptional level (5, 6, 7). IGFBP-1 is a 25-kDa IGFBP, distinct among the members of the IGFBP family in being acutely regulated by metabolic stimuli (8, 9, 10, 11). Studies in vivo or with cultured human liver explants suggest that the major regulatory influences on IGFBP-1 production are insulin, which is inhibitory (12), and hepatic substrate deprivation (13), which is stimulatory, acting through a cAMP-dependent mechanism. Food intake has a major suppressive effect on serum IGFBP-1 levels, leading to important fluctuations in serum IGFBP-1 in intermittently fed healthy subjects (9, 14, 15). In view of the continuous mode of feeding in prolonged critically ill patients, it is not surprising that serum IGFBP-1 concentrations are much more stable (3). Phosphoenolpyruvate carboxykinase (PEPCK) is the rate-limiting enzyme in gluconeogenesis, which is stimulated during critical illness (16, 17). The PEPCK gene promoter contains regulatory elements similar to that of the IGFBP-1 gene and shows similar transcriptional regulation with suppression by insulin and stimulation by cAMP (18). Increased glucose production by the liver is thought to be the major factor contributing to the hyperglycemia during critical illness (19).
Here, we report the effect of strict glycemic control with intensive insulin therapy on serum IGFBP-1 levels in the long-stay (>7 d) ICU patient group of the large insulin-in-ICU study (1), as well as its relationship to patient outcome. Furthermore, we have assessed the effects of intensive insulin therapy on hepatic IGFBP-1 and PEPCK gene expression in nonsurviving patients to obtain an indication of hepatic insulin responsiveness, pointing to changes in gluconeogenic activity.
Subjects and Methods
Subjects
This study was part of a large randomized controlled study on intensive insulin treatment in ICU patients (n = 1548), of which the major clinical outcomes have been published in detail elsewhere (1). In that study, all mechanically ventilated, adult patients admitted to a mainly surgical ICU were eligible for inclusion, after informed consent from the closest family member.
On ICU admission, patients were randomly assigned to either intensive or conventional insulin treatment. In the conventional group, continuous insulin infusion [50 IU Actrapid HM (Novo Nordisk A/S, Bagsvaerd, Denmark) in 50 ml of 0.9% NaCl using a Perfusor-FM pump (B. Braun, Melsungen, Germany)] was started only when the blood glucose level exceeded 11.9 mmol/liter (215 mg/dl) and adjusted to keep glycemia between 10 and 11.1 mmol/liter (180 and 200 mg/dl). In the intensive insulin group, insulin infusion was started when blood glucose levels exceeded 6.1 mmol/liter (110 mg/dl) and adjusted to maintain normoglycemia (4.4–6.1 mmol/liter or 80–110 mg/dl). Maximal insulin dose was arbitrarily set at 50 IU/h. At discharge from ICU, a conventional approach was adopted (glycemia 11.1 mmol/liter or 200 mg/dl). Whole blood glucose concentration was systematically measured every 1–4 h using the ABL700 analyzer (Radiometer Medical A/S, Copenhagen, Denmark). In addition, blood was systematically sampled upon admission and daily at 0600 h until ICU discharge or death. After clotting and centrifugation, serum was kept frozen at -80 C until assay. Furthermore, postmortem liver biopsies were taken from patients who died in ICU, immediately after they were declared dead. Samples were taken from the lower right quadrant of the liver, snap-frozen in liquid nitrogen, and stored at -80 C until analysis. The study protocol had been approved by the Institutional Review Board of the Catholic University of Leuven.
For the current analysis of the effect of intensive insulin treatment on serum IGFBP-1 concentrations and its prediction of ICU mortality, we selected all patients with an ICU stay of more than 7 d (n = 363). Because of the effect of intensive insulin treatment on morbidity, including shortening of ICU stay, this subgroup comprised 157 intensive insulin-treated patients and 206 patients receiving the conventional insulin approach. These patients had a median age of 66 yr, body mass index (BMI) of 26 kg/m2, and ICU stay of 16 d (Table 1
), and 27% of the patients were suffering from cardiac surgery-related critical illness. Because of the effect of insulin on ICU stay in cardiac surgery patients, in this selected group of long-stay patients, there were fewer patients after cardiac surgery in the intensive insulin group (22%) than in the conventionally treated group (32%; P = 0.04). Although the intensive insulin-treated group consisted of higher risk patients, mortality was 12% (n = 19) vs. 21% (n = 44) in the conventional insulin group (P = 0.02). Nonsurvivors had a median ICU stay of 22 d vs. 16 d in survivors (P = 0.03).
|
Serum IGFBP-1 concentration was measured on the admission day, then on d 1, 8, 15, 22, 29, and the last day of intensive care using an in-house enzyme-linked immunoassay specific for human IGFBP-1. To prepare biotinylated tracer, 10 µg IGFBP-1 was reacted with 1 mg/ml sulfo-NHS-LC-Biotin (Pierce Chemical Co., Rockford, IL) for 2 h at 2 C, and the product was separated from unreacted biotin by gel filtration. Goat antirabbit secondary antibody was precoated in excess on a 96-well microtiter plate and incubated at 22 C for 2 h; blocking buffer (100 mmol/liter NaH2PO4, 1% BSA, 0.01% Triton-X, 0.02% sodium azide) was added and then incubated at 37 C for 1 h. IGFBP-1 antiserum A2 (Ref. 20 ; 1:20,000 final concentration) was added to all except the nonspecific binding wells, then incubated at room temperature for 1 h. The microtiter plate was washed and blotted dry. Assays in duplicate contained 50 µl of serum samples and controls, diluted 1:5, and standards over the range 0, 0.03, 0.1, 0.3, 1, 3, and 10 ng/well. Fifty microliters of 1:100 dilution of IGFBP-1 biotin conjugate were subsequently added to all wells, and the plate was incubated at 4 C overnight. The plate was washed, and then 1:500 peroxidase-streptavidin complex in wash buffer was added and incubated at room temperature for 30 min. The plate was washed again, and 100 µl TMB Ultra Sensitive (Moss Inc., Pasadena, MD) were added to all wells. After a 10-min incubation at room temperature, the reaction was terminated with 2 mol/liter sulfuric acid, and the plate was read at 450 nm (EL312e, Bio-Tek Instruments, Inc. Winooski, VT).
The assay was optimized for the high levels of total IGFBP-1 found in critically ill patients, with 10 µl of serum routinely assayed in a final volume of 0.1 ml. Assay performance was analyzed for 50 successive assays. The mean absorbance (±SD) for the zero standard was 2.28 ± 0.09 absorbance units, and the ED50 was 0.52 ± 0.06 ng/0.1 ml assay volume. The limit of detection (B/Bo = 0.95 ± 0.02) was 0.03 ng IGFBP-1, equivalent to 3 ng/ml at the serum dilutions used. Assay precision was determined on three quality control samples run in duplicate in 50 assays. The within-assay coefficient of variation (CV) determined by ANOVA for the means of duplicates was 3.7% at 35 ng/ml, 3.7% at 56 ng/ml, and 3.9% at 222 ng/ml. At the same dose levels, the between-assay CV were 12.2%, 9.9%, and 9.9%, respectively.
Serum IGFBP-1 concentrations were not normally distributed, and thus log transformation was performed for data analysis. Data were analyzed by factorial and repeated measures ANOVA.
RNA isolation from liver biopsies and real-time PCR
Postmortem biopsies were taken from 74 of the 98 patients who died during the study (1), either acutely or after a long ICU stay. Insulin treatment and feeding were maintained during the moribund state of these patients until biopsies had been taken. Time between death and freezing of the liver samples was 30.5 ± 20.1 min. The samples were analyzed in a blinded fashion. Total RNA was isolated from freeze-clamped, whole liver tissue using the guanidine thiocyanate/cesium chloride method (21). Twenty micrograms of total RNA from all remaining liver samples were denatured and electrophoresed on a 1% agarose/formaldehyde gel, and the quality of the RNA was assessed by UV shadowing. Specific mRNAs were assayed by real-time RT-PCR (22) in the following way. Total RNA (5 µg) was denatured at 70 C, then annealed to 250 ng random hexamers (Invitrogen, Groningen, The Netherlands). Each sample was then reverse-transcribed using Superscript II (Life Technologies, Inc. Rockville, MD), 10 pmol deoxynucleotide triphosphates, 0.2 nmol dithiothreitol, 40 U RNAsin (Promega Corp., Madison, WI), and first strand buffer. Reactions were incubated for 12 min at 25 C, 50 min at 42 C, then 5 min at 70 C, and finally quenched on ice. All samples were reverse- transcribed simultaneously. Reactions lacking reverse transcriptase were also run to generate controls for assessment of genomic DNA contamination. A 1:100 dilution of the resultant cDNA was prepared, and 5 µl of this template was used in the real-time PCR protocol.
To create external standards for the quantitative PCR, gene-specific cDNAs were generated by RT-PCR (23). All primer pairs were designed to span an intron to avoid amplification of genomic DNA (Table 2). The fragments were then cloned into pGEM-T Easy (Promega Corp., Madison, WI), sequenced to confirm their identity, and quantified by spectrophotometry.
|
Primer sets were designed to amplify two separate intron-spanning regions (amplicons) of each of the IGFBP-1 and PEPCK transcripts, enabling us to assess the possible effects of RNA degradation that may have occurred during the postmortem period. In addition, gene expression was corrected for well-to-well loading variation by expressing data as a ratio to 18S rRNA, measured using the TaqMan Ribosomal RNA kit (PE Applied Biosystems, Foster City, CA). All samples were analyzed in duplicate, and percentage CV was calculated. Individual samples with a copy number CV greater than 20% were reanalyzed. The within-assay CV on the copy number quantification, determined by ANOVA for the means of duplicates at low, medium, and high expression levels was 20.5%, 15.4%, and 14.3%, respectively, for the IGFBP-1 exon 2–exon 3 primer set (BP1 Ex23); for BP1 Ex34: 18.2%, 11.3%, and 7.9%; for PEPCK Ex56: 23.7%, 10.7%, and 8.4%; for PEPCK Ex78: 13.7%, 8.4%, and 7.6%; for 18S rRNA: 40.4%, 22.3%, and 13.1%. All patient samples were analyzed in two runs. A separate run was performed for repeat samples. Because only a limited number of runs were performed, interassay precision was not formally assessed. Because of the wide range and lack of normal distribution in mRNA quantification, data were log-transformed before further analysis.
Results
Effect of strict maintenance of normoglycemia with insulin on serum IGFBP-1 concentrations in long-stay ICU patients
The two study groups, intensively insulin-treated and conventionally treated patients, were comparable for gender, age, BMI, history of diabetes, prevalence of hyperglycemia on admission, and severity of illness as reflected by on-admission APACHE-II score (Table 1
). In this long-stay ICU patient cohort, insulin treatment reduced ICU mortality, an effect that could not be attributed to randomization bias because the nonsurvivors in the intensive insulin group had higher APACHE-II scores on admission than the nonsurvivors in the conventional group [scores, 16 (range,12, 13, 14, 15, 16, 17, 18) vs. 12 (range,10, 11, 12, 13, 14, 15, 16, 17), respectively; P = 0.036], but equal APACHE-II scores on the day of death [24 (range,16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28) vs. 20 (range,16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28), respectively; P = 0.6] indicating that the intensively insulin-treated nonsurvivors were more severely ill upon admission. Furthermore, intensive insulin treatment reduced the occurrence of critical illness polyneuropathy, bacteremia, acute renal failure requiring dialysis, hyperbilirubinemia and inflammation, and reduced ICU stay, as previously reported (1).
Upon ICU admission, blood glucose concentrations were equal in both study groups [8.2 ± 3.2 mmol/liter (mean ± SD) in the conventional treatment group, and 8.1 ± 3.0 mmol/liter in the intensive insulin treatment group; P = 0.5], with hyperglycemia (>11 mmol/liter) present in only 13–14% of the patients. However, with nutrition started as soon as patients were hemodynamically stable, blood glucose levels peaked during the first 24 h of ICU stay to 10.7 ± 3.7 mmol/liter in the conventional treatment group and 9.9 ± 3.4 mmol/liter in the intensive insulin treatment group (P = 0.02). Within 24 h, blood glucose levels were controlled at a level of 9.2 ± 2.3 mmol/liter in the conventional treatment group and 6.0 ± 1.4 mmol/liter in the intensive insulin treatment group (P < 0.0001) and remained stable thereafter in both survivors and nonsurvivors (Fig. 1A). All intensive insulin-treated patients received exogenous insulin, and in 95% of these patients, insulin was infused from d 1 to the end of the ICU stay with a median daily dose of 69.1 IU. In contrast, only 52% of the conventionally treated patients received insulin any time during their ICU stay, with a median daily dose of only 0.4 IU. Unexpectedly, serum IGFBP-1 serum levels were unaffected by the different insulin treatments. Moreover, IGFBP-1 serum levels, measured on the last day before death or discharge from ICU were markedly elevated in the nonsurvivors as compared with the survivors (P < 0.0001), independent of gender (Fig. 1B).
|
). Three-factor ANOVA (for randomized insulin treatment assignment, outcome, and time) indicated that there was no direct effect of strict glycemic control with insulin on serum IGFBP-1 concentrations at any time during the treatment period, but that both outcome and time were significantly associated with serum IGFBP-1 concentrations (P < 0.0001). It also revealed an interaction between the effect of time and patient outcome (P < 0.0001) in the sense that serum IGFBP-1 concentrations decreased over time in the survivors but started to increase from d 8 onward in the nonsurvivors. Last-day serum concentrations of IGFBP-1 were lower than admission levels in the survivors (P < 0.0001) but higher than admission levels in the nonsurvivors (P = 0.026). This link between serum IGFBP-1 and death was clearly independent of the level of blood glucose control in the face of equal caloric intake, which was increased to normal levels over the first week of treatment (Fig. 2B). No differences in serum IGFBP-1 levels could be detected between patients who were exclusively parenterally fed and those receiving enteral feeding (data not shown). During the first 2 wk of intensive care, nonprotein caloric intake, corrected for body weight, was inversely correlated with serum IGFBP-1 levels. Among long-stay patients, the association between IGFBP-1 and caloric intake was: d 1, P = 0.0002, R = 0.20 (n = 356); d 8, P < 0.0001, R = 0.25 (n = 334); d 15, P = 0.015, R = 0.18 (n = 188); d 22, P = 0.81, R = 0.02 (n = 119); and d 29, P = 0.25, R = 0.13 (n = 86).
|
). Fifty-five patients in the conventional group and 32 patients in the intensive insulin treatment group were included in this retrograde analysis. Serum IGFBP-1 concentrations were significantly higher in nonsurvivors (n = 23) than survivors (n = 64) as long as 3 wk before death. This suggests that serum IGFBP-1 is potentially predictive of outcome 3 wk before death.
|
To assess which clinical factors might explain the relationship between an adverse outcome and elevated serum IGFBP-1 concentrations weeks before death, we performed stepwise (backward and forward) regression analysis between the log serum IGFBP-1 concentration on the last day —which best discriminated survivors from nonsurvivors —and biochemical markers of organ function and inflammation on the same day [APACHE-II score, serum total bilirubin concentration, prothrombin time (percentage), white blood cell count, thrombocyte count, C-reactive protein level, temperature, plasma urea and creatinine concentration]. Only APACHE-II, serum total bilirubin concentration, thrombocyte and white blood cell count were independently related to serum IGFBP-1, together accounting for 55% of the variance in serum IGFBP-1 (R = 0.74; R2 = 0.55).
There was a significant relationship between cause of death and serum IGFBP-1 concentration on the last day, as indicated by factorial ANOVA (Fig. 4). Last day serum IGFBP-1 concentrations were highest in patients who died of an acute cardiovascular collapse or multiple organ failure and sepsis compared with those who died from severe brain damage. The former two causes of death are accompanied by impaired tissue perfusion, whereas in the latter disease state, no such problem would be expected.
|
For each gene, levels of amplification of each amplicon were strongly associated, with correlation coefficients of 0.91 for the comparison of IGFBP-1 amplicons Ex23 and Ex34, and 0.97 for the comparison of PEPCK amplicons Ex56 and Ex78 (Fig. 5). Eight samples assessed as markedly degraded by visual inspection on gels were not processed for RT-PCR. Four samples with poor correlations between levels of IGFBP-1 amplicons Ex23 and Ex34 were also found to be partially degraded upon visual inspection of their RNA in gels and were excluded from the study. Three additional samples with calculated gene expression levels that lay more than 4 SD from the mean for both IGFBP-1 and PEPCK were also excluded, leaving 59 samples in the final analysis.
|
). The assumption that serum IGFBP-1 levels measured on the last day would reflect IGFBP-1 gene expression in postmortem liver samples was supported by the positive correlation between these measurements (R = 0.42 for serum IGFBP-1 vs. BP1 Ex23, and R = 0.46 for serum IGFBP-1 vs. BP1 Ex34, respectively; Fig. 6, B and C).
|
) or that of PEPCK Ex78 (data not shown), suggesting absence of the dominant regulatory effect of insulin as was also observed for the IGFBP-1 gene expression. These similarities were reflected in the positive correlation between PEPCK Ex78 and BP1 Ex34 gene expression (R = 0.58; P < 0.0001; Fig. 7B). Correlation between IGFBP-1 and PEPCK gene expression, as assessed using the other amplicons showed comparable associations (data not shown).
|
This study of 363 long-stay (>7 d) ICU patients, randomly allocated to either strict maintenance of normoglycemia with insulin or the conventional insulin strategy tolerating more elevated blood glucose levels, revealed that serum levels and hepatic gene expression of IGFBP-1 as well as PEPCK gene expression were not affected by intensive insulin therapy. This suggests that no major insulin effect on blood glucose regulation was occurring in the liver. However, a strong relationship between an elevated serum IGFBP-1 level and an adverse outcome was observed, significantly present weeks before death. The predictive value of a high serum IGFBP-1 concentration was similar to that of a high APACHE-II score.
IGFBP-1 is unique among the six IGFBPs in its acute regulation by metabolic status. The two dominant metabolic influences on IGFBP-1 gene expression appear to be insulin, which is suppressive, and the availability of hepatic substrate, with IGFBP-1 stimulated by substrate deprivation (7). Apart from its cellular effects on motility and adhesion, IGFBP-1 has a well defined role in the circulation as a regulator of free IGFs (i.e. IGFs not bound in high molecular weight complexes with the acid-labile subunit and IGFBP-3 or IGFBP-5; Refs. 11 and24).
An elevated serum IGFBP-1 level upon ICU admission has previously been suggested to be predictive of a catabolic state (25) and in the chronic phase of critical illness predictive of mortality (2, 3, 4). Because of the inhibitory effect of insulin and substrate supply, it was expected that subjects on intensive insulin therapy would have suppressed IGFBP-1 levels concomitant with their higher survival rate. Serum IGFBP-1 levels were clearly elevated in patients who ultimately died, but contrary to expectation, were completely unaffected by the mode of insulin treatment. The strong temporal dependence of IGFBP-1 levels during the course of critical illness was reflected by the fact that IGFBP-1 levels decreased over time in patients who recovered, whereas in those who ultimately died, IGFBP-1 increased steadily, despite normocaloric feeding being maintained. In the cohort of patients with an ICU stay of more than 4 wk, the levels of serum IGFBP-1 began to distinguish the nonsurvivor group from the survivor group at about 3 wk before death or discharge from ICU. This suggests that IGFBP-1 measurement might serve as an early predictive marker for mortality. However, the area under the curve of the ROC curves revealed that serum IGFBP-1 levels did not outperform the standard APACHE-II scoring system in predicting outcome in ICU patients. Although a strong link with mortality and severity of disease has previously been documented with biochemical markers such as lactate, IL-6, and C-reactive protein (26), they often have poor predictive values for outcome. Only indices, which combine different biochemical markers, like the endocrine index, appear to be more reliable, though still insufficient, predictors of mortality (27). Alternatively, to increase predictive power, biochemical markers could be included in broad physiological scoring systems like APACHE-II.
Although the intensive insulin treatment had a profound glucose lowering effect, indicating sensitivity of peripheral tissues to insulin action, the ineffectiveness of the insulin treatment on IGFBP-1, either in the serum or at the transcriptional level in the liver, suggests a marked hepatic insulin resistance during extended critical illness. In contrast, in other catabolic, insulinopenic conditions like type I diabetes (28), fasting (9), and anorexia nervosa (29), the inverse relationship between IGFBP-1 and insulin is clearly present. This study shows that, using serum IGFBP-1 as a marker of hepatic insulin sensitivity, long-term critically ill patients are markedly unresponsive to administered insulin. This contrasts with a more acute observational study in which serum IGFBP-1 levels did show an inverse relationship with serum insulin (30).
A dissociation between insulin and IGFBP-1 levels has also been described in liver cirrhosis, in which high serum IGFBP-1 levels were accompanied by elevated insulin levels (31, 32), despite markedly reduced hepatocyte number. Interestingly, in patients with chronic liver disease, serum IGFBP-1 levels were suppressed by an oral glucose tolerance test. This IGFBP-1-lowering effect was less in patients with more severe liver disease (32). In our study, caloric intake, corrected for body weight, suppressed IGFBP-1 levels during the first 2 wk of their ICU stay, independent of insulin treatment. Gradually over time, the regulatory potential of caloric input decreased. This phenomenon may underline the importance of preventing relative fasting during the acute phase of critical illness. The mechanism of loss of sensitivity to caloric intake during the extended phase of critical illness is less clear.
The paradoxical association of IGFBP-1 levels with poor outcome, while showing no regulation by insulin, suggests that the influence of other hormonal or metabolic factors may provide the link between IGFBP-1 and mortality. IGFBP-1 could be increased by cortisol (33), which is elevated in acute critical illness (34). However, during prolonged intensive care, serum cortisol levels decrease to normal levels (35, 36), and the link observed here between adverse outcome and high IGFBP-1 was most clearly present in the prolonged phase of critical illness. Alternatively, proinflammatory cytokines (IL-1, IL-6, and TNF-
) might be contributing factors. Similar to cortisol, these cytokines are markedly, but transiently, increased in the acute phase of critical illness, peaking at 2–3 h (37), whereas in the protracted critically ill patient, levels of some cytokines are either normal or only slightly elevated (38). This temporal pattern is quite different from that of serum IGFBP-1, which increases over weeks during the prolonged phase of critical illness. Another possible inductive influence on IGFBP-1 levels in critical illness may be cellular hypoxia (39, 40). Hypoxia could explain why patients who died from low organ perfusion, such as acute cardiovascular collapse and septic shock, had significantly raised serum IGFBP-1 concentrations. Examination of the molecular pathways affected by oxygen deprivation might provide further clarification of this possibility.
As another marker of hepatic insulin sensitivity, we measured hepatic gene expression of the rate-limiting gluconeogenic enzyme, PEPCK. In parallel with the lack of IGFBP-1 regulation by intensive insulin treatment, PEPCK gene expression similarly failed to show suppression by high-level insulin treatment. Like IGFBP-1, PEPCK is transcriptionally regulated by insulin in a dominant fashion (18). It has been reported that critical illness is marked by hyperglycemia due to both increased hepatic glucose production, through gluconeogenesis and glycogenolysis, and peripheral insulin resistance, causing decreased glucose uptake by peripheral tissues (19). The marked insensitivity of hepatic PEPCK mRNA levels to insulin treatment, despite its continued ability to lower serum glucose, calls into question the importance of gluconeogenesis relative to insulin-mediated glucose uptake by peripheral tissues. Furthermore, our data suggest that insulin resistance in extrahepatic tissues (particularly muscle and adipose tissue) during critical illness may be less substantial than previously thought. An alternative explanation for the discrepancy between the effects of insulin treatment on blood glucose levels and IGFBP-1 serum and gene expression levels may be the peripheral site of insulin administration. Intravenously administered insulin may not generate sufficiently high insulin levels in the portal vein to suppress the markers of hepatic insulin resistance, whereas it may directly increase glucose uptake in muscle and adipose tissue (41). However, because peripherally administered insulin is able to suppress elevated serum IGFBP-1 levels in diabetic patients (42), the site of insulin administration alone does not explain our findings.
In conclusion, we have shown that serum IGFBP-1 concentration during the course of critical illness is highly correlated with mortality, performing similarly to the APACHE-II score in predicting adverse outcome. IGFBP-1 may serve as a marker for hepatic insulin resistance, which in turn may play an important role in patient outcome, although the mechanism for this remains to be defined. A possible link between raised IGFBP-1 and adverse outcome might be organ hypoperfusion and the consequent inductive effect of tissue hypoxia on hepatic gene expression. The marked failure of insulin to inhibit IGFBP-1 and PEPCK gene expression, and perhaps gluconeogenesis, emphasizes the effect of critical illness on hepatic insulin resistance and points to the possible importance of peripheral insulin-mediated glucose uptake as the major glucoregulatory mechanism in these patients.
Acknowledgments
We acknowledge Pieter Wouters, Ilse Milants, Genny Gielens, Viviane Celis, Myriam Vandenbergh, and An Andries at Catholic University Leuven for sample handling and database management.
Footnotes
This work was supported by the Fund for Scientific Research-Flanders, Belgium [to D.M., Ph.D. scholarship, Aspirantenmandaat; and G.V.d.B. (G.0144.00) who is a Fundamental Clinical Research Investigator (G.3C05.95N) and who holds an unrestricted Novo Nordisk Research Chair]; and by the National Health and Medical Research Council of Australia (to P.J.D.D. and R.C.B.).
Abbreviations: BMI, Body mass index; CV, coefficient(s) of variation; ICU, intensive care unit; IGFBP, IGF binding protein; PEPCK, phosphoenolpyruvate carboxykinase; ROC, receiver-operating characteristic.
Received April 30, 2002.
Accepted August 25, 2002.
References
This article has been cited by other articles:
 |
|
 |
|
  B. Ellger, M. C. Richir, P. A. M. van Leeuwen, Y. Debaveye, L. Langouche, I. Vanhorebeek, T. Teerlink, and G. Van den Berghe Glycemic Control Modulates Arginine and Asymmetrical-Dimethylarginine Levels during Critical Illness by Preserving Dimethylarginine-Dimethylaminohydrolase Activity Endocrinology, June 1, 2008; 149(6): 3148 - 3157. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S B Catrina, R Rotarus, I R Botusan, M Coculescu, and K Brismar Desmopressin increases IGF-binding protein-1 in humans. Eur. J. Endocrinol., April 1, 2008; 158(4): 479 - 482. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Collier, L. A. Dossett, A. K. May, and J. J. Diaz Glucose Control and the Inflammatory Response Nutr Clin Pract, February 1, 2008; 23(1): 3 - 15. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Wallander, A. Norhammar, K. Malmberg, J. Ohrvik, L. Ryden, and K. Brismar IGF Binding Protein 1 Predicts Cardiovascular Morbidity and Mortality in Patients With Acute Myocardial Infarction and Type 2 Diabetes Diabetes Care, September 1, 2007; 30(9): 2343 - 2348. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. P. Sperandeo, P. Annunziata, A. Bozzato, P. Piccolo, L. Maiuri, M. D'Armiento, A. Ballabio, G. Corso, G. Andria, G. Borsani, et al. Slc7a7 disruption causes fetal growth retardation by downregulating Igf1 in the mouse model of lysinuric protein intolerance Am J Physiol Cell Physiol, July 1, 2007; 293(1): C191 - C198. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Langouche, I. Vanhorebeek, and G. Van den Berghe Glycaemic control in trauma patients, is there a role? Trauma, January 1, 2006; 8(1): 13 - 19. [Abstract] [PDF]
|
|
|
|
 |
|
 R. P. Peeters, P. J. Wouters, H. van Toor, E. Kaptein, T. J. Visser, and G. Van den Berghe Serum 3,3',5'-Triiodothyronine (rT3) and 3,5,3'-Triiodothyronine/rT3 Are Prognostic Markers in Critically Ill Patients and Are Associated with Postmortem Tissue Deiodinase Activities J. Clin. Endocrinol. Metab., August 1, 2005; 90(8): 4559 - 4565. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Basi, L. B. Pupim, E. M. Simmons, M. T. Sezer, Y. Shyr, S. Freedman, G. M. Chertow, R. L. Mehta, E. Paganini, J. Himmelfarb, et al. Insulin resistance in critically ill patients with acute renal failure Am J Physiol Renal Physiol, August 1, 2005; 289(2): F259 - F264. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Thorell, O. Rooyackers, P. Myrenfors, M. Soop, J. Nygren, and O. H. Ljungqvist Intensive Insulin Treatment in Critically Ill Trauma Patients Normalizes Glucose by Reducing Endogenous Glucose Production J. Clin. Endocrinol. Metab., November 1, 2004; 89(11): 5382 - 5386. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Z. T. Bloomgarden Inpatient Diabetes Control: Rationale Diabetes Care, August 1, 2004; 27(8): 2074 - 2080. [Full Text] [PDF]
|
|
|
|
|
|
D. Mesotten, P. J. Wouters, R. P. Peeters, K. V. Hardman, J. M. Holly, R. C. Baxter, and G. Van den Berghe Regulation of the Somatotropic Axis by Intensive Insulin Therapy during Protracted Critical Illness J. Clin. Endocrinol. Metab., July 1, 2004; 89(7): 3105 - 3113. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Mesotten, J. V. Swinnen, F. Vanderhoydonc, P. J. Wouters, and G. Van den Berghe Contribution of Circulating Lipids to the Improved Outcome of Critical Illness by Glycemic Control with Intensive Insulin Therapy J. Clin. Endocrinol. Metab., January 1, 2004; 89(1): 219 - 226. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Z. T. Bloomgarden Aspects of Blood Pressure, Lipid, and Glycemic Treatment Diabetes Care, January 1, 2004; 27(1): 264 - 269. [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
