high insulin-like growth factor binding protein 1 levels in cirrhosis: link with insulin resistance

High Insulin-like Growth Factor Binding Protein 1 Levels inCirrhosis: Link With Insulin Resistance

EHOUD SHMUELI,1 JOHN P. MIELL,2 MURRAY STEWART,1 K. GEORGE M. M. ALBERTI,1 AND CHRISTOPHER O. RECORD1

endocrine fashion, and their bioactivity is greatly modulatedHyperinsulinemic euglycemic clamps were performedby their binding proteins. Thus, GH, which reduces sensitiv-on six patients with compensated alcoholic cirrhosis andity to insulin, normally stimulates the secretion of IGF-I,on six normal comparison subjects. As in previous stud-which increases insulin sensitivity.ies, glucose uptake in the cirrhotic patients was only

Several roles have been suggested for the six IGF-binding21% of the comparison value. The cirrhotic patients hadproteins (IGFBPs) identified thus far. These include limitinghigh growth hormone (GH) and low insulin-like growththe bioactivity of the IGFs to protect against hypoglycemia,factor-I (IGF-I) levels, with low insulin-like growth fac-regulation of the stability or clearance rate of the IGFs, andtor–binding protein (IGFBP)-3 levels, but surprisinglymodulation of the action of the IGFs at a cellular level.2high IGFBP-1 levels (26.8 { 8.4 mgH vs. 3.2 { 0.2 mg/L, PIGFBP-3 is the most abundant in serum and binds 95% ofõ .001). The log IGFBP-1 level was inversely correlatedcirculating IGF-I and -II in a ternary complex that is highlywith the log insulin sensitivity (r Å 0.95). The clampsstable and probably serves to limit the bioavailability of thewere repeated with a somatostatin infusion to suppressIGFs. IGFBP-3 is produced by the liver and by most otherGH secretion. IGFBP-1 increased in both groups, espe-tissues in response to IGF-I and GH.2 IGFBP-1 is producedcially in the cirrhotic subjects. Insulin sensitivity in-mainly by the liver and is the only IGFBP that displays rapidcreased in the normal subjects but was unchanged inregulation with serum levels that vary rapidly in responsethe cirrhotic patients. Following GH treatment (0.13 U/to meals. Levels of IGFBP-1 are high in fetal life, fall precipi-kg/d for 5 days), the clamps were repeated. GH, IGF-I,tously at gestation, and then show a progressive age-relatedand IGFBP-3 levels were now similar in the two groups;decline in normal individuals. High levels have been foundIGFBP-1 levels decreased in the cirrhotic patients butin prolonged fasting, prolonged exercise, in anorexia nervosa,remained fivefold higher than the comparison valuein intensive care patients, in chronic renal failure, and in(10.6 { 3.7 vs. 2.1 { 0.4, P õ .05). Glucose uptake in theinsulin-dependent diabetes mellitus.4 IGFBP-1 levels arecirrhotic patients remained only 29% of the comparisonlowered by insulin. Low levels have been found in obesityvalue, but the change in their insulin sensitivity wasand polycystic ovary syndrome. IGFBP-1 is thought to limitinversely correlated with the change in their IGFBP-1the bioactivity of IGF-I by preventing its binding to cell mem-levels (r Å0.84). These results suggest an important rolebranes.4 Less is known about the other binding proteins.for IGFBP-1 in modulating insulin sensitivity in cirrho-IGFBP-2 may provide an intermediate level of regulation ofsis. (HEPATOLOGY 1996;24:127-133.)IGF availability and may be important under conditions inwhich there is insufficient IGFBP-3 to carry all availableGrowth hormone (GH) is essential for normal growth, butIGFs. It may be specifically induced by IGF-II.2its role in the adult remains uncertain. Its secretion is in-

The GH/IGF-I axis is deranged in cirrhosis5; GH levels arecreased by stress, starvation, and in diseases such as poorlyraised and increase further in response to insulin and glu-controlled diabetes mellitus, cirrhosis, and acromegaly. Incose, as opposed to the normal decrease. Despite the high GH,each of these conditions, high levels of GH are associatedIGF-I levels in cirrhosis are low. Cirrhotic patients exhibitwith insulin resistance. Indeed, infusions of GH in the shortmarked basal hyperinsulinemia and insulin resistance, butand long term have been shown to induce insulin resistance.1the relationship between the GH/IGF-I abnormalities andOne of the main effects of GH is to stimulate production ofthe metabolic abnormalities is unclear. It is possible that theinsulin-like growth factor-I (IGF-I). IGF-I and -II are polypep-imbalance between GH and IGF-I that exists in cirrhosistides with structural homology to proinsulin that regulateleads to insulin resistance. However, suppression of GH withproliferation and differentiation, but only IGF-I is dependentsomatostatin does not improve insulin sensitivity in cirrho-on GH.2 Like insulin, they exert insulin-like metabolic effectssis,6 and these data are now extended with IGFBP measure-and increase insulin sensitivity.3 Unlike insulin, IGFs arements. In an attempt to ‘‘equalize’’ GH and IGF-I levels be-produced by most tissues of the body and are abundant intween the cirrhotic and comparison subjects, both groupsthe circulation. They function in an autocrine, paracrine, andwere treated with GH. Thus, insulin sensitivity was mea-sured using hyperinsulinemic euglycemic clamps under threeconditions: 1) control clamps; 2) somatostatin infusion to sup-press GH secretion; and 3) following 5 days of treatment withAbbreviations: GH, growth hormone; IGF-I, insulin-like growth factor-I; IGFBP, insulin-

like growth factor–binding protein; NEFA, nonesterified fatty acid; SRIH,. GH. Six normal males and six patients with stable alcoholicFrom the 1Gastroenterology and Liver Unit, Royal Victoria Infirmary and Human Diabe- cirrhosis were studied.

tes and Metabolism Research Centre, Department of Medicine, University of Newcastle,Newcastle Upon Tyne, England, and 2Department of Medicine, Kings College School of PATIENTS AND METHODSMedicine, London, England.

Received July 12, 1995; accepted March 29, 1996. Subjects. Six male patients with stable biopsy-proven alcoholic cir-Supported by a grant from the Royal Victoria Infirmary Special Trustees and by an rhosis, previously demonstrated to be insulin resistant,5,6 and six

MRC research training fellowship to Dr. Shmueli.male comparison subjects were studied (Table 1). No patient hadAddress reprint requests to: E. Shmueli, M.D., Gastroenterology Unit, Northamptonascites, jaundice, or encephalopathy, and no patient was taking medi-General Hospital, Billing Road, Northampton NN1 5BD, England.cations known to affect glucose tolerance at the time of study. AllCopyright q 1996 by the American Association for the Study of Liver Diseases.

0270-9139/96/2401-0022$3.00/0 patients had endoscopic evidence of esophageal varices, and all had

127

AID Hepa 0021 / 5p10$$$401 06-21-96 11:16:16 hpta WBS: Hepatology

128 SHMUELI ET AL. HEPATOLOGY July 1996

TABLE 1. Clinical Characteristics of the Cirrhotic the glucose infusion bag (‘‘hot-infusion technique’’), according to the‘‘formula’’ devised and modified by Levy et al.7 to account for fastingand Normal Subjectsblood glucose:

Cirrhotic Subjects Normal Subjects(n Å 6) (n Å 6)

L Å F 1 glucose in bagHGO 1 IBW 1 (0.106 1 G / 0.349)

Age (yr) 54 { 3 30 { 1*L is the quantity of radioactivity added to the glucose infusion bagWeight (kg) 85.3 { 3.6 75.4 { 8.2(mCi), F is the isotope infusion rate during the basal period (mCi/BMI (kg/m2) 28.6 { 0.8 24.2 { 1.8min), HGO is the basal hepatic glucose production (assumed to beFFM (kg) 66.4 { 3.7 69.6 { 5.412 mmol/kg/min unless known from a previous study),5 IBW is theCreatinine (mmol/L) 65 { 12 83 { 2ideal body weight, and G is the basal blood glucose (mmol/L).Total protein (g/L) 69 { 2 63 { 1†

Insulin Sensitivity. Insulin sensitivity (SIP) was calculated ac-Albumin (g/L) 34 { 6 43 { 1cording to the formula:Bilirubin (mmol/L)) 16 { 4 10 { 2

ALP (U/L) 141 { 14 57 { 5*SIP Å

(Rd2/G2 0 (Rd1/G1)I2 0 I1

ALT (U/L) 34 { 6 31 { 3

NOTE. Values are means { SEM. Rd1 and Rd2 are the glucose disposal rates in the basal and hyperinsu-Abbreviations: BMI, basal metabolic index; FFM, fat-free mass; ALP, alka- linemic periods of the clamp, I1 and I2 are the respective serum insu-

line phosphatase; ALT, alanine transaminase. lin levels, and G1 and G2 are the blood glucose concentrations.8* P õ .01; †P õ .05. In the study with somatostatin, 250 mg/h was started at 0 minutes,

together with glucagon (0.4 ng/kg/min) and insulin (3-6 mU/m2/min)replacement.

Sampling and Analysis. Arterialized venous and deep venous sam-abstained from alcohol for at least 6 months. Blood alcohol levelsples were collected simultaneously at 0 minutes and at 10-minutemeasured randomly were consistently negative. The normal subjectsintervals between times 75 to 105 minutes and 270 to 300 minutes.were younger but were well matched for fat-free mass (Table 1). TheSamples were taken for the measurement of blood glucose, glycerol,purpose and risks of the study were explained to all subjects beforelactate, pyruvate, alanine, and plasma nonesterified fatty acidsconsent was obtained. The study protocol was approved by the New-(NEFA). Arterialized venous blood was also collected for measure-castle Health Authority and University of Newcastle Upon Tynement of insulin, C peptide, glucagon, IGF-1, GH, and tritiated glu-Joint Ethics Committee.cose.Protocol. Three euglycemic hyperinsulinemic clamps were per-

Blood glucose was measured by the glucose oxidase method usingformed on each subject—a control clamp and a clamp with somato-a glucose analyzer (Yellow Springs Instruments, Yellow Springs, OH;statin, in random order, followed at least 2 weeks later by 5 days ofinterstudy coefficient of variation, 2.2%). Blood for metabolites wastreatment with GH and a repeat clamp.collected into chilled perchloric acid (0.77 mol/L), and later assayedHuman biosynthetic GH (donated by Eli Lilly Industries, Basing-by fluorometric assay (interassay CV, 6%) adapted for centrifugalstoke, U.K.) was administered at a dosage of 0.13 U/kg/d by twice-analysis.9 Plasma NEFA were analyzed by centrifugal enzymaticdaily subcutaneous injections (3-4 milliliters per injection). All sub-analysis10 (interassay CV, 3%). Serum insulin and C peptide werejects received the GH for 5 days, and the clamp was performed onmeasured by radioimmunoassay11 (interassay CV, 6.5% and 6.2%,the sixth day, 12 hours after the last GH injection.respectively). Plasma glucagon was measured by radioimmunoas-The protocol for each clamp is illustrated in Fig. 1. Following ansay12 (interassay CV, 9.5%), and serum GH was measured by immu-overnight fast, a cannula was inserted into an antecubital vein fornoradiometric assay (IRMA kit North East Thames Region Immuno-infusions, and another was inserted retrogradely into an ipsilateralassay Unit, London, U.K.) (interassay CV, 3.5%). IGF-1 washand vein for arterialized venous blood sampling. Arterialization wasmeasured using a 125I RIA kit (Incstar Corporation, Stillwater, MN)achieved by warming in a hot-air hand box. A further cannula wason serum samples. The procedure involves extraction of the IGF-1inserted retrogradely into the contralateral antecubital vein for sam-from its binding proteins using sodium dodecyl sulfate-silica columnpling of blood draining mainly from the forearm muscle bed.extraction, methanol, and acetic acid. The extraction is followed byInfusions. Hepatic glucose production was measured using an iso-a double-antibody radioimmunoassay using rabbit anti–IGF-1 andtope technique. At 0 minutes, 17 mCi of high-pressure liquid chroma- 125I–IGF-1. The interassay CVs were 12.7%, 11.5%, and 8.3%, attography-purified [3-3H] glucose (New England Nuclear, Boston, MA)16, 35, and 61 nmol/L, respectively. The specific activity of tritiatedwas injected as an intravenous bolus, followed by a continuous infu-glucose was measured by deproteinization of a sample of plasmasion of 10 mCi/h. This dosage was reduced to 5 mCi/h at 150 minutes,with barium hydroxide, zinc sulphate, freeze drying, and liquid scin-which was further reduced to 2.5 mCi/h at 180 minutes, and contin-tillation counting. Serum levels of IGFBP-1 and -3 were measuredued until the end of the study. Tritiated glucose was also added toby specific radioimmunoassay.13 For IGFBP-1 assay, antiserum14

was used at a final dilution of 1:10,000, which bound approximately60% of iodinated tracer. Minimum detection limit of the assay was6 mg/L. The interassay CV at 55 mg/L was 6.2%, and the intra-assayat 35 mg/L was 4%. For IGFBP-3 the intra- and interassay CVs at3.5 mg/L were 4.2% and 11.4%, respectively.

Forearm Technique. Total forearm blood flow was measured witha mercury in rubber strain-gauge plethysmography after occlusionof the circulation to the hand, as previously described5:

Uptake of glucose Å blood flow 1 [arterialized venous 0

venous glucose concentration] Blood flow is expressed as milliters

per 100 mL of forearm per minute. Whole blood glucose is used.

Indirect Calorimetry. Substrate oxidation was calculated frommeasurements of oxygen uptake and carbon dioxide production usinga Deltatrack Indirect Calorimeter (Datex Instrumentation, Helsinki,Finland) with a ventilated hood, as previously described (Shmueli etal 1993). A timed urine collection was made for the duration of thestudy from which the urinary nitrogen excretion rate was determinedby the Kjeldahl method (Kjeltec Analyser, Perstorp analytical, Per-storp, Sweden).5

Fat-Free Mass. Fat-free mass was measured by bioelectrical im-pedance.15

FIG. 1. A schematic presentation of the protocol used. Statistical Analysis. Unless otherwise stated, all results are pre-


HEPATOLOGY Vol. 24, No. 1, 1996 SHMUELI ET AL. 129

TABLE 2. Arterialized Serum Concentrations of Insulin, C Peptide, in both groups, and a twofold increase in C-peptide levels inGlucagon, Whole Blood Glucose, and NEFA the cirrhotic patients (Table 2).

Glucagon. Basal glucagon levels were higher in the cir-Cirrhotic Subjects Normal Subjectsrhotic patients but fell to normal values during the clamp.

Blood glucose (mmol/L) Glucagon levels were not affected by the somatostatin infu-sion or by pretreatment with GH.Basal (control) 5.5 { 0.4 4.8 { 0.2

NEFA. Basal NEFA levels were higher in the cirrhotic pa-Basal (/SRIH) 5.8 { 0.3 4.7 { 0.2tients, and were inversely correlated with insulin sensitivityBasal (post-GH) 7.8 { 1.3 4.9 { 0.4*in each study (r Å 0.69, P Å .012, control study; r Å 0.80, P

Clamp (control) 5.3 { 0.3 4.7 { 0.2 Å .002, study with SRIH; and r Å 0.71, P Å .009, study post-Clamp (/SRIH) 5.1 { 0.3 4.9 { 0.4 GH). However, NEFA levels suppressed normally during theClamp (post-GH) 5.2 { 0.3 5.0 { 0.2 clamps and remained higher in the cirrhotic patients only inSerum insulin (mU/L) the clamp following GH treatment. SRIH had no effect on

NEFA levels in the cirrhotic subjects, but NEFA levels fellBasal (control) 25.1 { 3.9 5.2 { 0.4† in the normal subjects. GH pretreatment had no significantBasal (/SRIH) 16.9 { 2.4‡ 13.3 { 2.11‡

effect on basal NEFA levels.Basal (post-GH) 77.6 { 8.1‡ 16.5 { 5.0†‡GH and IGF-1. Basal GH levels tended to be higher in the

Clamp (control) 111.0 { 5.7 86.8 { 6.7* cirrhotic patients. During the control clamp, GH levels sup-Clamp (/SRIH) 80.2 { 14.8‡ 73.2 { 7.1 pressed in the comparison subjects but increased in five of theClamp (post-GH) 140.3 { 11.0 84.1 { 6.4§ six cirrhotic patients. Clamp IGF-1 levels were significantly

lower in the cirrhotic patients (Table 3). Somatostatin sup-C peptide (pmol/L)pressed GH levels in both groups.

Basal (control) 683 { 80 359 { 44§ One third of the subjects reported excessive fatigue duringBasal (/SRIH) 213 { 99‡ 51 { 46‡* treatment with GH. IGF-1 levels increased by variableBasal (post-GH) 1590 { 238‡ 582 { 150§

amounts in all subjects (Table 3). Basal GH levels were notClamp (control) 547 { 64 269 { 53§ significantly changed (the clamp was performed approxi-Clamp (/SRIH) 70 { 40‡ õ25‡* mately 12 hours after the last subcutaneous dose of GH, andClamp (post-GH) 1323 { 271‡ 346 { 122§ GH has a half-life of about 20 minutes). However, the para-Glucagon (pg/mL)

TABLE 3. Serum GH, IGF-I, and IGFBP-3Basal (control) 117 { 35 46 { 6*Basal (/SRIH) 105 { 35 32 { 6§

Cirrhotic Subjects Normal SubjectsBasal (post-GH) 109 { 23 49 { 7*

GH (mU/L)Clamp (control) 50 { 11 32 { 7Clamp (/SRIH) 40 { 5 32 { 6 Basal (control) 4.5 { 1.7 1.2 { 0.4Clamp (post-GH) 61 { 10 39 { 8 Basal (/SRIH) 2.0 { 1.5 õ0.5*

Basal (post-GH) 3.6 { 1.3 1.6 { 0.8NEFA (mmol/L)

Clamp (control) 6.1 { 0.4 0.5 { 0.4†Basal (control) 630 { 80 350 { 50*Clamp (/SRIH) 1.2 { 0.7 õ0.5Basal (/SRIH) 630 { 50 120 { 50*Clamp (post-GH) 1.8 { 0.6 2.1 { 1.0Basal (post-GH) 600 { 70 460 { 70

IGF-I (nmol/L)Clamp (control) 60 { 50 õ50 { 50Clamp (/SRIH) 80 { 30 õ50 { 50 Basal (control) 21 { 8 36 { 8Clamp (post-GH) 140 { 20 50 { 20§ Basal (/SRIH) 12 { 3 36 { 12*

Basal (post-GH) 41 { 13‡ 73 { 13‡NOTE. Values are expressed as means { SEM.* P õ .05 for cirrhotic vs. normal subjects. Clamp (control) 16 { 4 36 { 9*† P õ .001 for cirrhotic vs. normal subjects. Clamp (/SRIH) 23 { 6 44 { 6*‡ P õ .05 for differences within each group compared with control study. Clamp (post-GH) 46 { 14‡ 75 { 12‡§ P õ .01 for cirrhotic vs. normal subjects.

IGFBP-3 (mg/L)

Basal (control) 0.96 { 0.16 3.75 { 0.30§Basal (/SRIH) 1.21 { 0.42 3.41 { 0.33†sented as means { SEM. The significance of differences was testedBasal (post-GH) 4.07 { 0.90‡ 6.93 { 0.69‡by the Wilcoxon signed rank test (paired data) or the Mann-Whitney

U test (unpaired data). Clamp (control) 1.16 { 0.22 3.67 { 0.36§Clamp (/SRIH) 1.21 { 0.36 2.99 { 0.42†‡

RESULTS Clamp (post-GH) 4.13 { 0.86‡ 6.10 { 0.62‡

Blood Glucose. Fasting blood glucose levels were similar IGF-I:IGFBP-3 molar ratioin the patients with cirrhosis and the comparison subjects in

Basal (control) 0.6 { 0.3 0.2 { 0.1the control study, as well as in the study with somatostatin.Basal (/SRIH) 0.5 { 0.3 0.3 { 0.1Following GH treatment, blood glucose increased in five of Basal (post-GH) 0.3 { 0.1 0.3 { 0.1

the six cirrhotic patients and was allowed to fall back to theClamp (control) 0.4 { 0.1 0.3 { 0.1normal fasting level during the clamp. The mean CVs of bloodClamp (/SRIH) 0.9 { 0.4 0.5 { 0.2glucose measurements in the last hour of the clamp wereClamp (post-GH) 0.3 { 0.1 0.3 { 0.1below 3% in both groups (Table 2).

C-Peptide and Insulin Levels. In the control clamp, basal NOTE. Values are expressed as means { SEM.insulin levels were fivefold higher in the cirrhotic patients, † P õ .01 for cirrhotic vs. normal subjects.with higher C-peptide levels. Somatostatin considerably de- ‡ P õ .05 for differences within each group compared with control study.creased endogenous C-peptide secretion. Following GH treat- § P õ .001 for cirrhotic vs. normal subjects.

* P õ .05 for cirrhotic vs. normal subjects.ment, there was a threefold increase in basal insulin levels



FIG. 2. IGFBP-1 levels in the three studies. Values are means { SEM inmicrograms per liter. **P õ .01, *P õ .05 for cirrhotic vs. normal subjects; @P õ .05 for differences within each group compared with the control study.

doxical rise in GH that had characterized cirrhotic patientsin the control clamp did not occur, and GH levels were sup-pressed during the clamp.

IGFBP-1. IGFBP-1 levels tended to be higher in the cir-rhotic subjects in all the studies (Fig. 2). IGFBP-1 levels fellduring the clamps, but this fall was proportionally smallerin the cirrhotic patients such that the difference in IGFBP-1 levels was particularly prominent during the clamps. Inthe control study, there was a close inverse correlation be-tween the log IGFBP-1 levels in the cirrhotic subjects andtheir log insulin sensitivity (Fig. 3) (log IGFBP-1 Å 1.8 0 1log insulin sensitivity, r Å 0.95, P Å .003; and r Å 0.89,P õ .001 for all subjects). This relationship did not reachsignificance in the other studies (Fig. 4). Somatostatin signifi-cantly increased IGFBP-1 levels. Following GH treatment,IGFBP-1 levels were unchanged in the normal subjects but

FIG. 4. Insulin sensitivity (log mL/kg/min per mU/L) and IGFBP-1 (log mg/L) from all the subjects in each of the three studies. (A) Control Study; (B)Study with SRIH; (C) Study Post-GH.

diminished variably in the cirrhotic patients. The change inIGFBP-1 in the cirrhotic patients was inversely correlatedwith the change in insulin sensitivity (r Å 0.84, P õ .05) (rÅ .37 in the normal subjects).

IGFBP-3. IGFBP-3 levels were lower in the cirrhotic pa-tients, but the ratio of IGF-I to IGFBP-3 was similar in thetwo groups. Somatostatin increased this ratio in the normalsubjects only, implying greater availability of free IGF-I. TheGH treatment increased IGFBP-3 levels in all subjects, butthe IGF-I:IGFBP-3 ratio was unchanged (Table 3).

Glucose Uptake and Insulin Sensitivity. Total glucose up-FIG. 3. Insulin sensitivity (log mL/kg/min per mU/L) and IGFBP-1 (log mg/L) in the cirrhotic patients in the control study. take, forearm glucose uptake, and insulin sensitivity were



TABLE 4. Glucose Balanceconsiderably lower in the cirrhotic patients, with the predom-inant defect being impaired nonoxidative glucose disposal. Cirrhotic Subjects Normal SubjectsSomatostatin had no significant effect on these parameters

Hepatic glucose output (mg/kg/min)in the cirrhotic subjects but increased glucose uptake byabout 30% in the normal subjects. Glucose uptake and insulin

Basal (control) 1.69 { 0.16 2.09 { 0.1*sensitivity were not altered by GH treatment in the cirrhotic Basal (/SRIH) 1.42 { 0.28 2.21 { 0.26*subjects. In contrast, all measurements of glucose uptake Basal (post-GH) 1.96 { 0.13 2.42 { 0.38were diminished in all the normal subjects, forearm glucose

Clamp (control) 00.14 { 0.15 00.26 { 0.28uptake by about 30%, and insulin sensitivity by 25% (TableClamp (/SRIH) 0.14 { 0.15 00.05 { 0.024).Clamp (post-GH) 0.42 { 0.17† 0.88 { 0.38†Hepatic Glucose Output. Basal hepatic glucose output was

a little higher in the normal subjects, and suppressed nor-Clamp glucose requirement (mg/kg/min)mally in the control clamps and in the clamps with somato-

statin in both groups. Following GH, basal hepatic glucose Clamp (control) 3.29 { 0.56 9.42 { 1.14‡output increased in five of the six cirrhotic patients and in Clamp (/SRIH) 3.01 { 0.54 12.16 { 1.99†‡

Clamp (post-GH) 2.85 { 0.24 6.57 { 1.59*†four of the six normal subjects, despite the increase in basalinsulin levels. Moreover, hepatic glucose output failed to sup-press normally during the clamp in five of the six cirrhotic Forearm glucose uptake (mg/100 mL/min)patients and in all the normal subjects (Table 4).

Clamp (control) 0.27 { 0.04 1.22 { 0.42§Substrate Oxidation. Glucose oxidation was lower andClamp (/SRIH) 0.30 { 0.01 1.49 { 0.45§lipid oxidation higher in the cirrhotic patients during the Clamp (post-GH) 0.29 { 0.08 0.84 { 0.27§

control clamp. The differences were small and occurred inspite of the suppression of NEFAs. Glucose, lipid, and protein

Insulin sensitivity (mL/kg/min per mU/L)oxidation were unchanged by somatostatin. Following GHClamp (control) 24 { 8 114 { 20§treatment, lipid oxidation and basal metabolic rate increasedClamp (/SRIH) 35 { 10 189 { 29†§in the normal subjects. Protein oxidation was similar betweenClamp (post-GH) 25 { 5 85 { 27†§both groups and was unchanged by the treatment (Table 5).

NOTE. Values are expressed as means { SEM.DISCUSSION* P õ .05 for cirrhotic vs. normal subjects.

The control study confirmed that the cirrhotic patients † P õ .05 for differences within each group compared with control study.were characterized by a profound insulin resistance, associ- ‡ P õ .001 for cirrhotic vs. normal subjects.ated with low IGF-I, high GH, and high basal NEFA levels. § P õ .01 for cirrhotic vs. normal subjects.The principal new finding was high levels of IGFBP-1. Thismay be of particular significance because these patients wereboth hyperinsulinemic and older, compared with the compari- tance in cirrhosis predominantly affects peripheral (nonhe-

patic) tissues.5son subjects, and IGFBP-1 falls with age and in response toinsulin.4 Moreover, IGFBP-1 is not elevated in other hyperin- A possible consequence of peripheral insulin resistance is

that a greater proportion of substrate is available to be takensulinemic insulin–resistant states such as polycystic ovarysyndrome,4 and probably non–insulin-dependent diabetes up by the liver. We have previously demonstrated normal

splanchnic, but diminished peripheral, glucose uptake in cir-mellitus.16 It may therefore represent an important differencebetween hepatogenous diabetes and other insulin-resistant rhotic patients.19 This may be important when liver glycogen

reserves are depleted, or following liver damage. For exam-conditions. Another important point is the normal responseof IGF-I and IGFBP-3 to GH in cirrhosis. Previous findings ple, a large increase in circulating IGFBP-1 and in liver

IGFBP-1 messenger RNA occurs early following partial hepa-of a normal-high GH secretion with a low IGF-I have implieda GH resistance in cirrhosis; however, this is not the case tectomy in the rat, and may serve to divert available meta-

bolic resources toward the regenerating liver. It is also possi-because IGF-I and IGFBP-3 generation are not markedly im-paired. ble that IGFBP-1 has a role within the liver. Regenerating

hepatocytes and nonparenchymal liver cells possess IGF-IIGFBP-1 exerts its effects by specifically binding and mod-ulating the action of IGF-I. Each IGFBP-1 molecule binds a receptors and IGF-I stimulates lipocyte proliferation,20 al-

though its effect on collagen synthesis is unknown. The rapidsingle IGF molecule, and most in vitro studies suggest thatthe resultant complex inhibits IGF binding to the cell mem- increase in IGFBP-1 following partial hepatectomy19 may

suggest a role for IGFBP-1 in liver regeneration. One possibil-branes possibly by a conformational change in the IGF mole-cule.4 A number of in vivo observations suggest a counter- ity is that simultaneous secretion of IGF-I and IGFBP-1 from

hepatocytes results in high local concentrations of IGFBP-1/regulatory role for IGFBP-1 in glucose metabolism. IGFBP-1levels increase with fasting, or following hypoglycemia, and IGF-I complexes. IGFBP-1 has an Arg-Gly-Asp sequence that

may allow binding to local integrins.21 This may make IGF-fall after meals.4 IGFBP-1 has been shown to block the effectsof IGF-I in vitro and in vivo in rats. Indeed, administration I increasingly available to nonparenchymal liver cells stimu-

lating their proliferation and subsequent production of liverof IGFBP-1 on its own led to a transient rise in blood glucosein rats,17 suggesting that the free fraction of IGF-I has a matrix. It is possible that continued damage to hepatocytes

in chronic liver disease leads to continued overproductiontonic hypoglycemic effect that can be blocked. The very highIGFBP-1 levels in the insulin-resistant cirrhotic patients and of IGFBP-1, which, with locally produced IGF-I, acts as a

mitogenic stimulant to the lipocytes. Subsequent fibrosisthe strong inverse correlation between the log insulin sensi-tivity and the log IGFBP-1 levels implies a possible role for could itself cause further damage, leading to a vicious circle

and eventual cirrhosis. Conversely, the function of IGFBP-1IGFBP-1 in the etiology of the insulin resistance in cirrhosisby blocking the activity of free IGF-I. in the liver may actually be to inhibit the IGF-I–mediated

stimulation of nonparenchymal cells to prevent fibrosis inAlthough IGFBP-1 is produced by hepatocytes, only regen-erating hepatocytes possess IGF-I receptors.18 Normal hepa- response to injury. Cirrhosis may then result from insuffi-

cient IGFBP-1 production.tocytes would therefore be less affected by changes in thebioactivity of IGF-I. Hence, if IGFBP-1 is involved in glucose The Role of GH. The aim of the somatostatin part of the

study was to determine if suppression of GH over 5 hourscounter-regulation, this would explain why the insulin resis-



would improve insulin sensitivity. A surprising effect of so- peripheral tissues. Although GH has been shown to causeinsulin resistance both as a systemic infusion22-24 and as amatostatin was the increase in IGFBP-1 in both groups (by

43 mg/L in the cirrhotic, and 7.2 mg/L in the normal subjects). local infusion,25 its mechanism of action remains uncertain.One possibility is that GH induced lipolysis, resulting in sub-This increase may be explained by other effects of somato-

statin, including lower endogenous insulin secretion, as sug- strate competition between NEFA and glucose by way of theRandle cycle.26 There is some evidence for substrate competi-gested by the lower C-peptide levels, diminished hepatic

blood flow, or a combination of the two with reduced delivery tion in this study. Basal NEFA levels were higher in thecirrhotic patients, inversely correlated with insulin sensitiv-of metabolites, insulin, and other hormones to the liver. Fur-

thermore, somatostatin may itself stimulate IGFBP-1 pro- ity, and associated with higher lipid oxidation. Furthermore,although the increases in NEFA levels following GH treat-duction by hepatocytes, as demonstrated by the effect of its

long-acting analogue octreotide on human hepatoma cells.21 ment did not reach significance, lipid oxidation did increasein the normal subjects. However, during the clamps, NEFAThe large increase in IGFBP-1 in the cirrhotic patients

may explain why their insulin sensitivity failed to increase levels suppressed in both groups, implying that the differ-ences in the basal state were secondary to the insulin sensi-when GH was suppressed by somatostatin. Insulin sensitiv-

ity increased in the normal subjects, but the mechanism of tivity. Another possibility is that GH activates protein kinaseC,27 and a mechanism for protein kinase C modulation ofthis increase is uncertain. GH levels suppressed during the

control clamp and were not significantly lower with somato- insulin sensitivity has recently been suggested.28

It is possible that the insulin-resistance effect of GH maystatin. Although basal NEFA levels were lower with somato-statin, NEFA levels were completely suppressed during the be mitigated by locally produced IGF-I. If synthesis of IGF-

I occurs to a greater extent in more actively metabolizingclamps. IGF-I and IGFBP-3 levels were unchanged, butIGFBP-1 actually increased, raising doubts on the hypothesis tissues, as opposed to nonmetabolic or quiescent tissues, then

a mechanism would exist for local control of insulin sensitiv-that IGFBP-1 leads to insulin resistance. Nevertheless, itmay be that much greater increases in IGFBP-1 are required ity. Indeed, an increase in IGF-I messenger RNA has been

demonstrated in muscle following exercise.29 There is evento have an effect on insulin sensitivity.There is clear evidence that GH acutely impairs the action evidence that local production of IGFBP-3 may actually en-

hance the bioactivity of IGF-I.30 It is possible that the role ofof insulin on both the liver and peripheral tissues. In thenormal subjects, 5 days of treatment with GH diminished GH in the adult is to facilitate differences in IGF-I levels in

different tissues according to their metabolic requirements.glucose uptake and insulin sensitivity by 30%, and impairedsuppression of hepatic glucose output during the clamp. Al- In the cirrhotic patients, GH increased IGF-I and IGFBP-

3 to levels similar to the normal subjects. Insulin and C-though the calculated insulin sensitivity in the cirrhotic sub-jects was unchanged, basal insulin levels were dramatically peptide levels were also increased, and this is probably part

of the explanation for the fall in IGFBP-1 levels. Higher IGF-increased, implying gross insulin resistance. IGF-I levels in-creased in both groups, but this increase was balanced by an I levels may also have contributed, because IGF-I inhibits

gene expression of IGFBP-1 in HepG2 human hepatoma cells,increase in IGFBP-3 such that the IGF-I:IGFBP-3 ratio wasunchanged, implying no change in the level of free IGF-I. The which have abundant IGF-I receptors,31 and inhibits IGFBP-

1 expression from human decidual cells.32 Insulin sensitivityfall in insulin sensitivity cannot be attributed to IGFBP-1levels, which were unchanged in the normal subjects and fell increased in three of the cirrhotic patients, and decreased in

three. It is interesting that the change in insulin sensitivityto a varying extent in the cirrhotic patients. It must thereforebe attributed to a direct effect of GH on the liver and on correlated with the change in IGFBP-1, and this is further

evidence that IGFBP-1 is at least partly responsible for theinsulin resistance of cirrhosis.

In conclusion, a number of factors may contribute to theTABLE 5. Substrate Disposalinsulin resistance of cirrhosis, including persistent periph-

Cirrhotic Subjects Normal Subjects eral hyperinsulinemia through impaired hepatic insulin deg-radation, and high GH levels. However, the strikingly highGlucose oxidation (mg/kg/min)IGFBP-1 levels demonstrated in this study, and the correla-

Clamp (control) 2.02 { 0.16 3.42 { 0.52* tion between this parameter and insulin resistance, suggestClamp (/SRIH) 1.84 { 0.11 3.67 { 0.66* that it is a major factor. It is proposed that IGFBP-1 is theClamp (post-GH) 1.74 { 0.18 3.34 { 0.47* signal from the liver to the periphery denoting an increased

hepatic demand for available substrates. By diminishing theNonoxidative glucose disposal (mg/kg/min) bioactivity of IGF-I, apparent peripheral insulin resistance

is manifest, with the result that the proportion of availableClamp (control) 1.23 { 0.45 6.00 { 0.73†Clamp (/SRIH) 1.34 { 0.48 8.49 { 1.54‡§ substrates taken up by the liver is increased. Continued pro-Clamp (post-GH) 1.40 { 0.33 5.38 { 1.28‡ duction of IGFBP-1 in liver disease may explain the associ-

ated insulin resistance.Lipid oxidation (mg/kg/min)

Acknowledgment: The authors are grateful to Sisters Ma-Clamp (control) 0.44 { 0.07 0.16 { 0.09* vis Brown and Margaret Miller for their help and encourage-Clamp (/SRIH) 0.44 { 0.06 0.10 { 0.14* ment during the study, and to Linda Ashworth and her staffClamp (post-GH) 0.68 { 0.05 0.46 { 0.05*§

for technical assistance. Eli Lilly Industries are thanked forthe donation of growth hormone.

Protein oxidation (mg/kg/min)

Clamp (control) 0.35 { 0.06 0.38 { 0.07 REFERENCESClamp (/SRIH) 0.42 { 0.10 0.59 { 0.06

1. Møller N, Jørgensen JOL, Schmitz O, Moller J, Christiansen JS, AlbertiClamp (post-GH) 0.28 { 0.04 0.38 { 0.03KGMM, Orskov H, et al. Effects of a growth hormone pulse on total andforearm substrate fluxes in humans. Am J Physiol 1990;258:E86-E91.NOTE. Values are expressed as means { SEM. 2. Cohick WS, Clemmons DR. The insulin-like growth factors. Annu Rev

* P õ .05 for cirrhotic vs. normal subjects. Physiol 1993;55:131-53.† P õ .001 for cirrhotic vs. normal subjects. 3. Zenobi PD, Graf S, Ursprung H, Froesch ER. Effects of insulin-like growth‡ P õ .01 for cirrhotic vs. normal subjects. factor on glucose tolerance, insulin levels, and insulin secretion. J Clin

Invest 1992;89:1908-1913.§ P õ .05 for differences within each group compared with control study.



4. Lee PDK, Conover CA, Powell DR. Regulation and function of insulin-like hepatoma, and normal, regenerating, and fetal rat liver. J Clin Invest1988;81:976-981.growth factor-binding protein-1. Proc Soc Exper Biol Med 1993;204:4-29.

5. Shmueli E, Walker M, Alberti KGMM, Record CO. Normal splanchnic 19. Pinzani M, Abboud HE, Aron DC. Secretion of insulin-like growth factor-I and binding proteins by rat liver fat-storing cells: regulatory role ofbut impaired peripheral insulin stimulated glucose uptake in cirrhosis.

HEPATOLOGY 1993;18:86-95. platelet derived growth factor. Endocrinology 1990;127:2343-2349.20. Mohn KL, Melby AE, Tewari DS, Laz TM, Taub R. The gene encoding6. Shmueli E, Stewart M, Alberti KGMM, Record CO. Growth hormone, IGF-

I, and insulin resistance in cirrhosis. HEPATOLOGY 1994;19:322-328. insulin-like growth factor-binding protein 1 is rapidly and highly inducedin regenerating liver. Mol Cell Biol 1991;11:1393-1401.7. Levy JC, Brown G, Matthews DR, Turner RC. Hepatic glucose output in

humans measured with labelled glucose to reduce negative errors. Am J 21. Lee J, Greenbaum L, Haber BA, Nagle D, Lee V, Miles V, Mohn KL, etal. Structure and localization of the IGFBP-1 gene and its expressionPhysiol 1989;257:E531-E540.

8. Ader M, Bergman RN. Insulin sensitivity in the intact organism. Baillieres during liver regeneration. HEPATOLOGY 1994;19:656-665.22. Ren SG, Ezzat S, Melmed S, Braunstein GL. Somatostatin analog inducesClin Endocrinol Metab 1987;1:879-910.

9. Harrison J, Hodson AW, Skillen AW, Stappenbeck R, Agius L, Alberti insulin-like growth factor binding protein-1 (IGFBP-1) expression in hu-man hepatoma cells. Endocrinology 1992;131:2479-2481.KGMM. Blood glucose, lactate, pyruvate, glycerol, 3-hydroxybutyrate and

acetoacetate measurements in man using a centrifugal analyser with a 23. Bratusch-Marrain PR, Smith D, DeFronzo RA. The effect of growth hor-mone on glucose metabolism and insulin secretion in man. J Clin Endocri-fluorimetric attachment. J Clin Chem Clin Biochem 1988;26:141-146.

10. Knox DP, Jones DG. Automated enzymatic determination of plasma free nol Metab 1982;55:973-982.24. Rizza RA, Mandarino LJ, Gerich JE. Effects of growth hormone on insulinfatty acids by centrifugal analysis. J Automatic Chem 1984;6:152-154.

11. Soeldner JS, Slone D. Critical variables in the radioimmunoassay of serum action in man. Diabetes 1982;31:663-669.25. Fryburg DA, Louard RJ, Gerow KE, Gelfand RA, Barrett EJ. Growthinsulin using the double-antibody technique. Diabetes 1965;14:771-779.

12. Ørskov H, Thomsen HG, Yde H. Wick chromatography for rapid and reli- hormone stimulates muscle protein synthesis and antagonizes insulin’santiproteolytic action in humans. Diabetes 1992;41:424-429.able immunoassay of insulin, glucagon and growth hormone. Nature 1968;

219:193-195. 26. Randle PJ, Garland PB, Hales CN, Newsholme EA. The glucose fatty acidcycle. Lancet 1963;i:785-789.13. Miell JP, Taylor AM, Zini M, Maheshwari H, Ross RJM, Valcavi R. Effects

of hypothyroidism and hyperthyroidism on insulin-like growth factors 27. Slootweg MC, De Groot RP, Herrmann-Erlee MPN, Koornneep I, KruijerW, Kramer YK. Growth hormone induces expression of c-jun and jun-B(IGFs) and growth hormone (GH) and IGFBPs. J Clin Endocrinol Metab

1993;76:950-955. oncogenes and employs protein kinase C signal transduction for the induc-tion of the c-fos oncogene expression. J Mol Endocrinol 1991;6:179-188.14. Taylor AM, Dunger DB, Preece MA, Holly JMP, Smith CP, Wass JAH,

Patel S, Tate VE, et al. The growth hormone independent insulin-like 28. Shmueli E, Alberti KGMM, Record CO. Hypothesis—diacylglycerol andprotein kinase C signalling: a mechanism for insulin resistance. J Interngrowth factor binding protein, BP-28 is associated with serum insulin-

like growth factor-I inhibitory bioactivity in adolescent insulin dependent Med 1993;234:397-400.29. Yan Z, Biggs RB, Booth FW. Insulin-like growth factor immunoreactivitydiabetics. Clin Endo 1990;32:229-239.

15. Zillikens MC, van den Berg JWO, Wilson JHP, Rietveld T, Swart GR. The increases in muscle after acute eccentric contractions. J Appl Physiol 1993;74:410-414.validity of bioelectrical impedance analysis in estimating total body water

in patients with cirrhosis. J Hepatol 1992;16:59-65. 30. Ernst M, Rodan GA. Increased activity of insulin-like growth factor (IGF)in osteoblastic cells in the presence of growth hormone (GH): positive16. Bang P, Brismar K, Rosenfeld RG, Hall K. Fasting affects serum insulin-

like growth factors (IGFs) and IGF-binding proteins differently in patients correlation with the presence of the GH induced IGF-binding protein BP-3. Endocrinology 1990;127:807-814.with noninsulin-dependent diabetes mellitus versus healthy nonobese and

obese subjects. J Clin Endocrinol Metab 1994;78:960-967. 31. Lee PDQ, Suwanichkul A, Depaolis LA, Snuggs MB, Morris SL, PowellDR, et al. IGF suppression of IGFBP-1 production—evidence for mediation17. Lewit MS, Denyer GS, Cooney GJ, Baxter RC. Insulin-like growth factor–

binding protein-1 modulates blood glucose levels. Endocrinology 1991;129: by the type 1 IGF receptor. Regul Pept 1993;48:199-206.32. Thrailkill KM, Clemmons DR, Busby WH, Handwerger S. Differential2254-2256.

18. Caro JF, Poulos J, Ittoop O, Pories WJ, Flickinger EG, Sinha MK. Insulin- regulation of IGFBP secretion from human decidual cells by IGF-I. J ClinInvest 1990;86:878-883.like growth factor I binding in hepatocytes from human liver, human


high insulin-like growth factor binding protein 1 levels in cirrhosis: link with insulin resistance

Documents