Effect of Congestive Heart Failure on the Insulin-LikeGrowth Factor-1 System

Asif Anwar, MD, Jean-Michel Gaspoz, MD, Sandro Pampallona, PhD,Aliya Aftab Zahid, MD, Philippe Sigaud, RN, Claude Pichard, MD, Marijke Brink, PhD,

and Patrick Delafontaine, MD

Congestive heart failure (CHF) is characterized bymultiple neurohumoral alterations including stim-

ulation of the renin-angiotensin system.1 We recentlydemonstrated that angiotensin II infusion in the ratproduces weight loss,2 similar to the cachexia thatcharacterizes advanced CHF,3 and decreases circulat-ing and tissue insulin-like growth factor-1 (IGF-1)levels. IGF-1 is ubiquitously expressed, stimulates cellgrowth, differentiation, and anabolism, and its bioac-tivity is modulated by IGF-binding proteins (IGFBPs)including the main circulating carrier IGFBP-3.4 Asmall fraction of IGF-1 circulates as the biologicallyactive free form.5 Our animal findings provided a linkbetween the renin-angiotensin system and IGF-1, andprompted us to conduct a prospective study to deter-mine levels of circulating IGF-1, IGFBP-3 and freeIGF-1 in patients with CHF treated with and withoutangiotensin-converting enzyme inhibitors.

• • •We studied patients admitted for new-onset and/or

worsening CHF in New York Heart Associationclasses II to IV and age-matched controls. Clinical,laboratory, and echocardiographic data were prospec-tively collected, and admission IGF-1, free IGF-1, andIGFBP-3 were measured using a commercially avail-able kit (Diagnostic Systems Laboratories Inc., Web-ster, Texas). A mini-nutritional assessment score wasperformed,6 and body fat mass, fat-free mass, and totalbody water were assessed using bioelectrical imped-ance analysis.7

Statistical analyses were performed with the statis-tical software Stata (Stata Corporation, College Sta-tion, Texas). Clinical, echocardiographic, nutritional,and laboratory data were compared using chi-squaretests for categorical data and the Kruskal-Wallis testfor continuous data. Linear regression was used toinvestigate the simultaneous effect of factors onIGF-1, IGFBP-3, and free IGF-1. Backward stepwiseregression was used with variables leaving the modelif their p value was �0.1 and entering the model iftheir p value was �0.05. The only term for casenessthat remained in all final models was the one that

identified controls (indicating no difference betweenpatients with systolic or diastolic dysfunction); for thepurposes of presentation, controls have been taken asreference. Results of the multivariate regression arereported in terms of the coefficient, their correspond-ing SE, p value, and 95% confidence interval. All pvalues are for 2-sided tests.

Sixty-nine of 233 screened patients were excludedfrom the analysis because they did not meet the entrycriteria. Patients with symptoms of CHF and an ejec-tion fraction �50% were included in the systolicdysfunction group, and patients with an ejection frac-tion �50% were included in the diastolic dysfunctiongroup. Patient characteristics are listed in Table 1. Thecontrol group had a lower incidence of coronary arterydisease, hypertension, and peripheral vascular disease.

Univariate analysis (Figure 1) showed that totalIGF-1 levels were not statistically different among the3 groups (p � 0.15). The pooled IGF-1 value in the 2CHF groups was 83.2 � 4.9 (mean � SEM) versus94.6 � 5.4 ng/ml for controls (p � 0.13). However,using multivariate analysis (Table 2), patients withCHF had lower adjusted total IGF-1 than controls,with a mean reduction of 20.5 � 9.2 ng/ml (p �0.027). This decrease was almost completely bluntedby the use of angiotensin-converting enzyme inhibi-tors, because patients with these medications had amean adjusted increase of 18.1 � 9.1 ng/ml of IGF-1(p � 0.049). Men had higher circulating IGF-1 thanwomen, and total IGF-1 decreased with age as previ-ously described. Surprisingly, patients with CHF tak-ing molsidomine had significantly lower mean ad-justed total IGF-1.

Patients with both systolic and diastolic CHF had amarked decrease in circulating IGFBP-3; mean IG-FBP-3 levels in all patients with CHF was 1,916 � 83ng/ml (mean � SEM), representing a 36% decreasecompared with controls (p �0.0001; Figure 1). Mul-tivariate analysis (Table 2) confirmed that patientswith systolic or diastolic CHF had a decrease of ad-justed mean circulating IGFBP-3. Additionally, pa-tients with systolic or diastolic CHF had markedlyhigher circulating free IGF-1 levels than controls (Ta-ble 2). Overall, mean free IGF-1 in all patients withCHF was 9.89 � 0.6 ng/ml (mean � SEM), repre-senting a 74% increase compared with controls (p�0.0001 for comparison between all groups or allsubjects with CHF and controls). Multivariate analysis(Table 2) indicated that no other variable, besides thepresence of CHF, significantly affected circulatingfree IGF-I levels.

Left ventricular mass was increased in patients

From the Division of Cardiology, Medical Clinic II, Department ofInternal Medicine, and Division of Nutrition, University Hospital, Ge-neva, Switzerland; ForMed, Statistics for Medecine, Evolene, Valais,Switzerland; and Division of Cardiovascular Diseases, The Universityof Kansas Medical Center, Kansas City, Kansas. Dr. Delafontaine’saddress is: Division of Cardiovascular Diseases, 1001 Eaton Building,The University of Kansas Medical Center, 3901 Rainbow Boulevard,Kansas City, Kansas 66160. E-mail: pdelafontaine@kumc.edu.Manuscript received December 27, 2001; revised manuscript re-ceived and accepted August 1, 2002.

1402 ©2002 by Excerpta Medica, Inc. All rights reserved. 0002-9149/02/$–see front matterThe American Journal of Cardiology Vol. 90 December 15, 2002 PII S0002-9149(02)02885-0

with systolic CHF (Table 1). No correlation was foundbetween left ventricular mass or ejection fraction andtotal IGF-1, free IGF-1, IGFBP-3, or nutritional status.The diet score, as assessed by the mini-nutritionalassessment, was higher in controls (23 � 0.25) than inpatients with systolic (20 � 0.3) or diastolic (20 �0.6) CHF (p �0.001). Body mass index was slightlyhigher in patients with HF, and reached statisticalsignificance only in women (Table 1). Bioimpedanceanalysis showed that height2/resistance (an index offat-free mass), fat-free mass, and total body waterwere not significantly different between patients withHF and controls. Univariate and multivariate analysesdid not reveal any correlation between the IGF-1system and nutritional or anthropometric analysis.

• • •In the present study, we demonstrated several

novel observations in patients with CHF. Serum levelsof total IGF-1 and of its main circulating carrier pro-tein, IGFBP-3, were decreased irrespective of themechanism and the severity of left ventricular failure.In contrast, free IGF-1 was greatly increased. Further-more, the decrease in total IGF-1 was not observed

when patients were treated with angiotensin-convert-ing enzyme inhibitors. This was not the case forchanges in IGFBP-3 and free IGF-1. Because of themultiple effects of IGF-1 on cell growth, survival, andmetabolism (including insulin-like effects on glucose,protein, and lipid metabolism), these modifications ofthe IGF-1 system may be important in the understand-ing of the clinical syndrome of CHF.

To our knowledge, this is the first report of changesin the IGF-1 system in a prospectively studied popu-lation of nonselected patients hospitalized for CHF.Broglio et al8 recently reported lower total IGF-1levels, without a change in IGFBP-3, in selected pa-tients with HF waiting for heart transplantation; freeIGF-1 was not measured. Osterziel et al9 demonstratedthat in patients with dilated cardiomyopathy, totalIGF-1 levels correlated positively to systolic function.In contrast, Anker et al10 reported no overall change intotal IGF-1 levels in 53 men with stabilized CHFcompared with 16 controls. A subgroup of 16 patientswith cachectic CHF was found to have higher growthhormone levels, without a change in total IGF-1 com-pared with patients with noncachectic CHF. In a sub-

TABLE 1 Clinical, Echocardiographic, and Laboratory Characteristics

Systolic CHF(n � 78)

Diastolic CHF(n � 25)

Control(n � 60) p Value*

Age (yrs) 75 � 1 77 � 2 80 � 1 0.120Men/women 46/32 7/18 27/33 0.019Systolic blood pressure (mm Hg) 138 � 3 140 � 5 141 � 2 0.28Diastolic blood pressure (mm Hg) 82 � 2 81 � 3 81 � 1 0.51Heart rate (beats/min) 86 � 2 82 � 4 74 � 11 �0.0001Atrial fibrillation 18 (23%) 7 (28%) 2 (3%) 0.001Coronary artery disease 47 (60%) 1 (4%) 2 (3%) �0.001Valvular heart disease 8 (10%) 2 (8%) 0 �0.001Dilated cardiomyopathy 13 (17%) 0 0 �0.001Peripheral vascular disease 20 (26%) 4 (16%) 1 (2%) �0.001Systemic hypertension 54 (69%) 17 (68%) 0 �0.001Non–insulin-dependent diabetes mellitus 17 (22%) 5 (20%) 4 (7%) 0.036Cigarette smoker 33 (42%) 7 (28%) 21 (35%) 0.40Medical treatment

Molsidomine 14 (18%) 1 (4%) 1 (2%) 0.002Nitrates 44 (56%) 12 (48%) 1 (2%) �0.001Diuretics 73 (94%) 21 (84%) 2 (3%) �0.001� blockers 12 (15%) 2 (8%) 0 0.002Digoxin 42 (54%) 8 (32%) 3 (5%) �0.001Angiotensin receptor antagonists 10 (13%) 3 (12%) 0% 0.006Angiotensin-converting enzyme inhibitors 41 (53%) 15 (60%) 1 (2%) �0.001Calcium blockers 9 (12%) 1 (4%) 1 (2%) 0.056Hematocrit (%) 41 � 1 38 � 1 42 � 1 0.002Total protein (g/L) 70 � 1 71 � 1 71 � 1 0.8Albumin (g/L) 34 � 1 36 � 2 37 � 1 0.016Creatinine (�mol/L) 116 � 4 108 � 7 89 � 1 �0.001C-reactive protein (mg/L) 46 � 7 52 � 18 6 � 1 �0.001

EchocardiographyLeft ventricular mass (g)

Men 175 � 10 141 � 16 113 � 3 �0.001Women 181 � 18 143 � 18 112 � 6 0.005

Fractional shortening (%) 19 � 1 38 � 3 38 � 1 �0.001Ejection fraction (%) 32 � 1 61 � 1 61 � 1 �0.001Body mass index (kg/m2)

Men 26 � 1 23 � 2 24 � 1 0.06Women 25 � 1 27 � 1 24 � 1 0.02

*Probability by Kruskal-Wallis test, comparing diastolic CHF versus systolic CHF versus control. Otherwise all the probabilities are calculated by Fisher’s exact testcomparing diastolic CHF versus systolic CHF with control.

Results are expressed as mean � 1 SEM.

BRIEF REPORTS 1403

sequent study, the same investigators11 found a sig-nificant decrease in circulating IGF-1 in 21 patientswith cachectic CHF compared with patients with non-cachectic CHF.11 Niebauer et al12 reported decreasedskeletal muscle cross-sectional area and strength inpatients with CHF with low IGF-1 levels.

Based on our findings, we postulate that one of themajor regulators of circulating IGF-1 in HF is anincreased level of angiotensin II. Thus, in our patients,the decrease in total IGF-1 was blocked in the pres-ence of angiotensin-converting enzyme inhibitor ther-apy. This is consistent with a report from Corbalan etal13 and with our animal data demonstrating that an-giotensin II infusion reduces hepatic IGF-1 synthesisand circulating total IGF-1.14 Alternatively, increasedlevels of tumor necrosis factor-� in advanced HF mayalso contribute to some of these changes becausetumor necrosis factor-� infusion in a rat model15 de-creased circulating IGF-1. Although angiotensin IIinfusion in animals produces weight loss,3 in ourpopulation of nonselected patients admitted for CHF,

we could not show any catabolic ef-fects as assessed by body mass indexor bioelectric impedance analysis.One can speculate that the decreasein total IGF-1 may precede subse-quent changes in metabolic balancein these patients, and subsequentmuscle wasting and body weightloss. Also, bioelectric impedancemeasurements may not be adequateto assess total body water in acuteedematous states.16

The changes in IGFBP-3 and freeIGF-1 were profound and unexpected.IGFBP-3 is often co-regulated with to-tal IGF-1, as in acromegaly.17 The reg-ulation of free IGF-1 is not completelyunderstood, but in acromegaly it in-creases together with total IGF-1.5Free IGF-1 may be an important reg-ulator of anabolic actions of IGF-1.18

The increase in free IGF-1 in our pa-tients may have been related to severalmechanisms. The ternary IGFBP-3/ac-id-labile subunit/IGF-1 complex doesnot cross the endothelium and regu-lates IGF-1 bioavailability by servingas a reservoir for circulating IGF-1.Thus, the proportionally greater de-crease in IGFBP-3 (compared with to-tal IGF-1) may lead to increased freeIGF-1. This may represent a compen-satory change in response to the de-crease in total IGF-1. However, thechange was not corrected by angioten-sin-converting enzyme inhibitors, sug-gesting that it is independent of thechange in total IGF-1. Importantly,low free IGF-119 has been associatedwith decreased self-reported quality ofhealth in an elderly population. Con-

versely, free IGF-1 has been reported to be higher inhealthy persons aged �70 years, with a concomitantdecrease in total IGF-1, IGFBP-3, free androgen, andfree estrogen.19

The increase in free IGF-1 may also reflect changesin IGF-1 turnover, stability, or tissue diffusion, poten-tially linked to alterations in IGF-1 receptors on tissueIGFBPs. Thus, in the angiotensin II-infused animal,14

there was a strong decrease in skeletal IGFBP-3 andIGFBP-5 levels. Thus, one can speculate that in-creased free IGF-1 is a compensatory change in re-sponse to tissue IGF-1 depletion or decreased tissueIGF-1 binding capacity. If tissue IGF-1 storage capac-ity is decreased in CHF, then a further increase in freeIGF-1 induced by growth hormone treatment mayhave limited beneficial effects. Several preliminarytrials20 of growth hormone in CHF have yieldedequivocal results. The striking decrease in IGFBP-3 inpatients with CHF was not blocked by angiotensin-converting enzyme inhibitors, suggesting other mech-anisms (e.g., changes in IGFBP-3 protease activity).4

FIGURE 1. Univariate analysis of IGF-I (A), IGFBP-3 (B), and free IGF-I (C). Probabilityby Kruskal-Wallis test comparing diastolic CHF versus systolic CHF versus control. *p� 0.053; †p <0.0001; ‡p <0.0001.

TABLE 2 Multivariate Analysis of IGF-1, IGFBP-3, and Free IGF-1 (ng/ml)

Coeff. SE p Value 95% CI

IGF-1CHF �21 9 0.027 �39 �2Angiotensin-converting enzyme

inhibitor18 9 0.049 0.1 36

Age/yr �1 0.4 0.004 �2 �0.4Men 26 11 0.02 4 48Molsidomine �27 12 0.032 �52 �2

IGFBP-3CHF �1175 169 �0.0001 �1508 �841Age/yr �34 9 �0.0001 �52 �15

Free IGF-1CHF 4 1 �0.0001 2 6

CI � confidence interval; Coeff. � coefficient.

1404 THE AMERICAN JOURNAL OF CARDIOLOGY� VOL. 90 DECEMBER 15, 2002

Our study shows that in an elderly population ofpatients hospitalized for CHF, there was a signifi-cant decrease in total IGF-1, a profound decreasein IGFBP-3, and a marked increase in circulatingfree IGF-1. The decrease in total IGF-1 was notpresent in patients who received angiotensin-con-verting enzyme inhibitors.

1. Levine TB, Francis GS, Goldsmith SR, Simon AB, Cohn JN. Activity of thesympathetic nervous system and renin-angiotensin system assessed by plasmahormone levels and their relation to hemodynamic abnormalities in congestiveheart failure. Am J Cardiol 1982;49:1659–1666.2. Brink M, Wellen J, Delafontaine P. Angiotensin II causes weight loss anddecreases circulating insulin-like growth factor I in rats through a pressor-independent mechanism. J Clin Invest 1996;97:2509–2516.3. Anker SD, Ponikowski P, Varney S, Chua TP, Clark AL, Webb-Peploe KM,Harrington D, Kox WJ, Poole-Wilson PA, Coats AJ. Wasting as independent riskfactor for mortality in chronic heart failure. Lancet 1997;349:1050–1053.4. Rajaram S, Baylink DJ, Mohan S. Insulin-like growth factor-binding proteinsin serum and other biological fluids: regulation and functions. Endocrinol Rev1997;18:801–831.5. Frystyk J, Skjaerbaek C, Dinesen B, Orskov H. Free insulin-like growth factors(IGF-I and IGF-II) in human serum. FEBS Lett 1994;348:185–191.6. Vellas B, Guigoz Y, Garry PJ, Nourhashemi F, Bennahum D, Lauque S,Albarede JL. The Mini Nutritional Assessment (MNA) and its use in grading thenutritional state of elderly patients (see comments). Nutrition 1999;15:116–122.7. Pichard C, Kyle UG, Bracco D, Slosman DO, Morabia A, Schutz Y. Referencevalues of fat-free and fat masses by bioelectrical impedance analysis in 3393healthy subjects. Nutrition 2000;16:245–254.8. Broglio F, Fubini A, Morello M, Arvat E, Aimaretti G, Gianotti L, Boghen MF,Deghenghi R, Mangiardi L, Ghigo E. Activity of GH/IGF-I axis in patients withdilated cardiomyopathy (see comments). Clin Endocrinol 1999;50:417–430.9. Osterziel KJ, Ranke MB, Strohm O, Dietz R. The somatotrophic system inpatients with dilated cardiomyopathy: relation of insulin-like growth factor-1 andits alterations during growth hormone therapy to cardiac function. Clin Endocri-nol 2000;53:61–68.

10. Anker SD, Chua TP, Ponikowski P, Harrington D, Swan JW, Kox WJ,Poole-Wilson PA, Coats AJ. Hormonal changes and catabolic/anabolic imbalancein chronic heart failure and their importance for cardiac cachexia (see comments).Circulation 1997;96:526–534.11. Anker SD, Volterrani M, Pflaum CD, Strasburger CJ, Osterziel KJ, DoehnerW, Ranke MB, Poole-Wilson PA, Giustina A, Dietz R, Coats AJ. Acquiredgrowth hormone resistance in patients with chronic heart failure: implications fortherapy with growth hormone. J Am Coll Cardiol 2001;38:443–452.12. Niebauer J, Pflaum CD, Clark AL, Strasburger CJ, Hooper J, Poole-WilsonPA, Coats AJ, Anker SD. Deficient insulin-like growth factor I in chronic heartfailure predicts altered body composition, anabolic deficiency, cytokine andneurohormonal activation. J Am Coll Cardiol 1998;32:393–397.13. Corbalan R, Acevedo M, Godoy I, Jalil J, Campusano C, Klassen J. Enalaprilrestores depressed circulating insulin-like growth factor 1 in patients with chronicheart failure. J Card Fail 1998;4:115–119.14. Brink M, Price SR, Chrast J, Bailey JL, Anwar A, Mitch WE, DelafontaineP. Angiotensin II induces skeletal muscle wasting through enhanced proteindegradation and down-regulates autocrine insulin-like growth factor I. Endocri-nology 2001;142:1489–1496.15. Fan J, Char D, Bagby GJ, Gelato MC, Lang CH. Regulation of insulin-likegrowth factor-I (IGF-I) and IGF-binding proteins by tumor necrosis factor. Am JPhysiol 1995;269:R1204–R1212.16. Gudivaka R, Schoeller DA, Kushner RF, Bolt MJ. Single- and multifrequencymodels for bioelectrical impedance analysis of body water compartments. J ApplPhysiol 1999;87:1087–1096.17. van der Lely AJ, de Herder WW, Janssen JA, Lamberts SW. Acromegaly: thesignificance of serum total and free IGF-I and IGF-binding protein-3 in diagnosis(discussion S15–S16). J Endocrinol 1997;155(suppl 1):S9–S13.18. Skjaerbaek C, Frystyk J, Grofte T, Flyvbjerg A, Lewitt MS, Baxter RC,Orskov H. Serum free insulin-like growth factor-I is dose-dependently decreasedby methylprednisolone and related to body weight changes in rats. Growth HormIGF Res 1999;9:74–80.19. Janssen JA, Stolk RP, Pols HA, Grobbee DE, Lamberts SW. Serum free andtotal insulin-like growth factor-I, insulin-like growth factor binding protein-1 andinsulin-like growth factor binding protein-3 levels in healthy elderly individuals.Relation to self-reported quality of health and disability. Gerontology 1998;44:277–280.20. Volterrani M, Manelli F, Cicoira M, Lorusso R, Giustina A. Role of growthhormone in chronic heart failure. Therapeutic implications. Drugs 2000;60:711–719.

Type of Liver Dysfunction in Heart Failure and ItsRelation to the Severity of Tricuspid Regurgitation

George T. Lau, MBBS, Hiok C. Tan, MBChB, and Leonard Kritharides, PhD

Congestive cardiac hepatopathy is characterized bythe triad of clinical heart failure (especially right-

sided cardiac failure), abnormal liver function tests(LFTs) and the exclusion of alternate causes of liverdysfunction, which usually involves infective and au-toimmune serology and hepatic imaging. Surprisingly,the description of the typical profile of LFT abnormal-ity in heart failure is poorly defined and apparentlycontradictory. The Oxford Textbook of Medicine1 de-scribes elevation predominantly of transaminases asbeing typical. A recent review2 and several otherinvestigators3–5 describe a mixed pattern, whereas oth-ers6 describe a predominantly cholestatic profile or nopattern.7–9 Defining the LFT profile typical of heartfailure may avoid unnecessary hepatic investigations.Early recognition of hepatic complications may alsowarrant more intensive treatment of heart failure and

specific treatment of causative factors. This studyaims to (1) identify the characteristic pattern of LFTabnormality in heart failure, and (2) identify associ-ated cardiac factors to elucidate the processes medi-ating hepatic impairment.

• • •The records of all patients admitted to the cardiol-

ogy unit of a tertiary referral hospital with a diagnosisof left, right, or congestive heart failure (as recordedby the diagnosis-related group classification) during a16-month period were examined. All patients who hada transthoracic echocardiogram recorded during ad-mission and LFTs within 5 days of the echocardio-gram were identified. Patients were excluded if therewas evidence of acute myocardial infarction on elec-trocardiography or creatine kinase (which frequentlycauses hemodynamic shock and raised transami-nases), a confounding cause of increased LFT results,an echocardiogram of insufficient quality for analysis,or if a previous admission had already been includedin the study. Of 142 patients identified, 32 were ex-cluded (15 had an acute myocardial infarction, 1 hadhepatobiliary disease, 5 had an alcohol consumption

From the Department of Cardiology, Concord Repatriation GeneralHospital, Concord, Australia. Dr. Lau’s address is: Department ofCardiology, Concord Repatriation General Hospital, Hospital Road,Concord, New South Wales 2139, Australia. E-mail: george.lau@email.cs.nsw.gov.au. Manuscript received May 5, 2002; re-vised manuscript received and accepted August 9, 2002.

1405©2002 by Excerpta Medica, Inc. All rights reserved. 0002-9149/02/$–see front matterThe American Journal of Cardiology Vol. 90 December 15, 2002 PII S0002-9149(02)02886-2