The favourable effects of growth hormone (GH) substitution ... - Nature

0 downloads 0 Views 142KB Size Report
in symptom-free adults with hypopituitarism.2 In. 1989, Jщrgensen et al3 and Salomon et al4 reported that treatment of GHD adults with growth hormone.
International Journal of Obesity (1998) 22, 529 ±536 ß 1998 Stockton Press All rights reserved 0307±0565/98 $12.00 http://www.stockton-press.co.uk/ijo

The favourable effects of growth hormone (GH) substitution on hypercholesterolaemia in GH-de®cient adults are not associated with concomitant reductions in adiposity. A 12 month placebo-controlled study N Vahl1, JOL Jùrgensen1,5, TB Hansen2, IB Klausen3, A-G Jurik4, C Hagen2 and JS Christiansen1,5 1 Medical Department M (Endocrinology and Diabetes), Aarhus Kommunehospital, Aarhus C; 2Department of Endocrinology, Odense University Hospital, Odense; 3Department of Cardiology, Aarhus Amtssygehus, Aarhus; 4Department of Radiology, Aarhus Kommunehospital, Aarhus C; and 5Center for Clinical Pharmacology, Aarhus University Hospital, Aarhus, Denmark.

OBJECTIVE: To investigate whether the changes in lipoproteins following growth hormone (GH) substitution in GH de®cient (GHD) adults are determined by the concomitant changes in body composition and physical ®tness in a controlled long-term study. DESIGN: A randomized, double-blind, placebo-controlled trial with GH (2 IU/m2) or placebo given for 12 months. SUBJECTS: Twenty-seven patients (18 male, 9 female, aged 21±61 y) with adult onset GH de®ciency. Comparisons were made with age- and gender-matched healthy adults. MEASUREMENTS: Serum triglycerides (TG) and lipoproteins, body composition (Dual-Energy X-ray Absorptiometry  and computerized tomography), and exercise capacity (VO2-max measured by bicycle ergometry) were measured at baseline and after 12 months. RESULTS: Baseline values of total cholesterol, low-density lipoprotein (LDL) and serum triglycerides were signi®cantly higher in GHD adults compared to normal subjects (P < 0.001, P < 0.001, P ˆ 0.004, respectively) whereas no difference in high-density lipoprotein (HDL) or lipoprotein (a) (Lp(a)) was found. After one year of GH treatment total cholesterol decreased signi®cantly (P ˆ 0.02). Serum LDL decreased after GH and increased after placebo, but the difference in delta values was not signi®cant (P ˆ 0.12). Serum HDL and TG concentrations were unchanged. Lp(a) increased but not signi®cantly. Serum total and LDL cholesterol remained signi®cantly elevated after one year of GH treatment. Signi®cant reductions in total and visceral adiposity, and improved exercise capacity were also recorded after GH treatment. In normal subjects, serum total cholesterol and TG correlated positively with age, subcutaneous fat and  intraabdominal fat, and negatively with VO2-max. Serum LDL correlated positively with age. In GHD patients, baseline values of serum TG correlated positively with subcutaneous fat and serum insulin. During treatment, no signi®cant correlations were found between the changes in lipoproteins and in body composition. CONCLUSION: The cholesterol lowering effect of GH is not determined by the concomitant decrease in adiposity, which supports the concept of a direct effect of GH on lipoprotein metabolism. Keywords: growth hormone de®cient adults; lipid pro®le; body composition; exercise capacity; long-term GH treatment

Introduction Growth hormone de®ciency (GHD) in adults is associated with increased vascular morbidity and mortality.1 An increased prevalence of atherosclerosis in the carotid arteries has been detected by ultrasonography in symptom-free adults with hypopituitarism.2 In 1989, Jùrgensen et al 3 and Salomon et al 4 reported

Correspondence: Nina Vahl, MD, Medical Department M, Aarhus Kommunehospital, DK-8000 C, Denmark. Received 28 August 1997; revised 26 November 1997; accepted 20 January 1998

that treatment of GHD adults with growth hormone (GH) decreased fat mass and increased lean body mass. Subsequently, it was revealed that visceral fat mass was increased in GHD patients, and that GH treatment caused a 30% reduction of visceral fat mass and a smaller reduction in subcutaneous fat.5,6 Visceral fat is a well known risk factor for cardiovascular disease and death,7 but it is not known whether the increased prevalence of ischaemic heart disease (IHD) in GHD is secondary to the excess visceral fat mass. Increased serum concentrations of cholesterol, lowdensity lipoprotein (LDL) cholesterol and apolipoprotein B (apo B), and decreased concentration of highdensity lipoprotein (HDL) cholesterol are also well recognized risk factors for cardiovascular disease,8

Lipoproteins in GH-de®cient adults N Vahl et al

530

and lipoprotein (a) (Lp(a)) is an alleged independent risk factor for atherosclerotic disease.6 Most 2; 9ÿ12 but not all13 studies agree that GHD patients exhibit hypercholesterolaemia caused by an increased concentration of LDL cholesterol. Serum HDL cholesterol is reported in the normal range,2,10 decreased11,13,14 or even increased12 in GHD patients. Serum triglycerides (TG) are unchanged10,12 or increased.9,13 The main structural protein of LDL, apo B, is found to be increased in the patients.10,11 A recent study5 found that levels of Lp(a) were unchanged compared to controls matched for gender but not for age. Replacement of GHD with GH promotes favourable changes in lipoproteins in most studies although increments in Lp(a) are reported in some,15,16 but not all, trials.17±19 Whether the changes in lipoproteins are secondary to changes in body composition or physical ®tness, remain to be examined. Studies so far have been of either short duration or inadequately controlled. Furthermore, most studies have failed to include reference data from an appropriate population of healthy adults. Since the favourable effects of GH on visceral adiposity and lipoproteins are important rationales for GH substitution in adults, it is essential to obtain robust long-term data. In the present study, we evaluated the impact of 12 months placebo-controlled GH therapy in GHD adults on lipoproteins, body composition and physical ®tness as compared to a control group of untreated healthy subjects.

Subjects and methods Subjects

Twenty-seven patients (18 male, 9 female; mean age  s.e.m. 44.5  1.8 y, range 21± 61 y) with adult onset GHD of at least one year duration, were included. All patients had pituitary disease and the diagnosis was ultimately based on a GH stimulation test (insulin tolerance test or arginine infusion). The mean peak serum GH value was 2.2  0.5 mg=L (0 ±8.0 mg=L). None of the patients had previously been treated with GH. Additional replacement therapy had been stable for at least six months prior to the study, and continued unchanged throughout the study (Table 1). Characteristics of the patients are given in Table 1. For comparison, 27 healthy age- and gendermatched subjects were investigated. All subjects gave informed written consent. The study was approved by the regional Ethical Committee and the National Board of Health, and conducted according to the Declaration of Helsinki and the guidelines for Good Clinical Practice. Study design

The study had a randomized, double-blind, placebocontrolled design of 12 months duration. Patients were strati®ed prior to randomization, according to age, gender and Cushing vs non-Cushing. The patients had been diagnosed as GHD for at least 1 year (mean 5.7 y, range 1±25 y) prior to the start of the study.

Table 1 Patient characteristics Patient number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

Gender

Age (y)

Diagnosis

Hormone replacement

GH/Pl

F F F F M M M M M M M M M M F F F M M F M M M M M M F

49 46 54 51 21 38 52 53 52 52 46 55 44 49 51 53 48 31 33 49 56 32 43 61 35 21 38

Pituitary adenoma Epidermoid cyst Pituitary adenoma Pituitary adenoma Pituitary apoplexy Pituitary adenoma Pituitary adenoma Pituitary adenoma Pituitary adenoma Pituitary adenoma Craniopharyngioma Pituitary adenoma Pituitary adenoma Pituitary adenoma Pituitary adenoma Pituitary adenoma Meningioma Paraganglioma Pituitary adenoma Pituitary adenoma Pituitary adenoma Craniopharyngioma Pituitary adenoma Pituitary adenoma Pituitary adenoma Germinoma Pituitary adenoma

H, L D, E, H, L E, H, L None H, L, T D, H, L, T H None None H, L, T H, L, T None None T None H H, L H, L, T H, L, T None L, T H, L, T H, L H H, L, T H H, E

Pl GH Pl GH Pl GH Pl Pl Pl GH GH GH Pl GH Pl Pl GH GH Pl GH GH Pl GH GH Pl Pl Pl

M ˆ male; F ˆ female; D ˆ desmopressin; E ˆ estradiol; H ˆ hydrocortisone; L ˆ levothyroxine; T ˆ testosterone; GH ˆ growth hormone; Pl ˆ placebo.

Lipoproteins in GH-de®cient adults N Vahl et al

Additional hormone replacement had been stable and unchanged for at least six months. Fourteen (9 male, 5 female) subjects received GH and 13 (8 male, 5 female) subjects received placebo. Growth hormone (Norditropin, Novo Nordisk A=S, Gentofte, Denmark) or placebo (Novo Nordisk A=S, Gentofte, Denmark) was administered as daily subcutaneous injections at bed-time. The starting dose of GH was 0.5 IU=m2=d which was increased stepwise by 25% at 14 d intervals until a maintenance dosage of 2 IU=m2=d was reached. In the case of side effects, the dose was reduced by 25% and further reductions were allowed if complaints persisted. Body composition, exercise capacity and lipid pro®le were measured at baseline and after 12 months. Hormone analyses

Serum insulin-like growth factor I (IGF-I) was measured by a non-competitive time resolved immuno¯uorometric assay,20 and serum insulin by conventional RIA. Serum IGF binding protein-1 (IGFBP-1) was measured by a commercial ELISA (Medix Biochemica, Kainiainen, Finland). Plasma cholesterol, HDL cholesterol and triglycerides (TG) were measured by standard enzymatic kits. LDL cholesterol was calculated by the Friedewald's equation21 in subjects with TG < 4.5 mmol=l. Lp(a) was measured with a two-site immunoradiometric assay (Pharmacia, Uppsala, Sweden) as described previously.22 A standard curve was constructed for each run, and all samples were analysed in duplicate. Our laboratory participated in an international Lp(a) standardization program, and the intra- and interassay coef®cients of variation were 2.2% and 3.1%, respectively. Body composition and physical ®tness

The amount of intraabdominal (visceral) fat and the muscle=fat ratio of the mid-thigh region were evaluated by computed tomography (CT) with a Somatom Plus-S scanner as earlier described.23 Percentage body fat and total lean body mass (LBM) were measured by Dual-Energy X-ray Absorptiometry (DEXA) using a Hologic QDR-2000 densitometer (Hologic, Waltham, MA). In addition, body mass index (BMI, weight (kg)=height (m)2) and waist to hip ratio (WHR) were measured. In the  normal controls, maximal oxygen consumption (VO2-max, ml=min=kg) was determined on a bicycle ergometer according to the Ê strand protocol.24 In patients, exercise capacity (kJ) A was measured on a bicycle ergometer with an initial workload of 2.9 kJ=min increased by 2.9 kJ=min every 3 min until exhaustion. Statistical analyses

Differences between controls and patients in serum IGF-I, body composition and lipid pro®le were assessed by Students' t-test for paired data. Treatment

effects were evaluated by comparing delta values (baseline±12 months) in the GH vs placebo group by Students' t-test for unpaired data. Pearson's product moment correlation was used to measure the strength of association between the variables, and multiple regression analysis was used to determine whether any of the variables were able to predict the changes in lipid pro®le. Analyses were made on log-transformed data when they were not normally distributed. Data are given as mean  s.e.m. for normally distributed data and otherwise as medians and 25 ±75 percentiles. Statistical signi®cance was assumed for P < 0.05. If multiple comparisons were made a protected P value of < 0.01 was employed. The randomization code was not made available for statistical analysis until all data had been entered twice and cross-checked.

Results Data on body composition, physical ®tness and serum IGF-I levels have been described in detail earlier.25 Brie¯y, GHD patients at baseline had signi®cantly greater amounts of intraabdominal and subcutaneous abdominal fat than normal controls. After GH treatment, visceral and subcutaneous abdominal fat decreased by 25% and 17%, respectively, to levels no longer different from the control group. Total body fat as assessed by DEXA was signi®cantly higher in GHD adults at baseline, compared to controls, but normalized after GH treatment. Exercise capacity (kJ) increased signi®cantly after GH treatment. Serum IGF-I concentrations were low at baseline (156  8 mg=l (controls) vs 86  12 mg=l (GHD baseline), P < 0.01)) and rose above normal after GH substitution (275  29 mg=l (GH 12 months), P < 0.001). No signi®cant differences in serum TG or lipids were recorded at baseline between the GH and the placebo group. As shown in Table 2, GHD adults at baseline had signi®cantly higher levels of serum total cholesterol (P < 0.001), LDL cholesterol (P < 0.001), and TG (P ˆ 0.004), whereas concentrations of HDL cholesterol and Lp(a) in serum were within the normal range. Baseline serum levels of insulin were not different from the control group (11.2  1.9 mU=l (controls) vs 9.6  1.0 mU=l (GHD baseline), NS), whereas serum IGFBP-1 was higher in the patients (3.1  0.3 (controls) vs 6.3  1.1 mg=l (GHD baseline), P ˆ 0.04). Treatment with GH for 12 months (Table 3) caused a signi®cant decrease in serum total cholesterol (P ˆ 0.02). Serum LDL cholesterol decreased insigni®cantly in the GH treated group and increased in the placebo group, but the difference in delta values only approached signi®cance (P ˆ 0.12). The increase in serum Lp(a) was signi®cant in the GH group

531

Lipoproteins in GH-de®cient adults N Vahl et al

532

Table 2 Mean  s.e.m. values of lipoproteins in growth hormone (GH) de®cient adults (GHDA) before and after 12 months treatment with GH vs normal age- and gender-matched controls. For triglycerides (TG) and lipoprotein (a) (Lp(a)) medians and 25 ±75 percentiles are given Controls Cholesterol (mmol/L) HDL cholesterol (mmol/L) LDL cholesterol (mmol/L) TG (mmol/L) Lp(a) (mg/dl)

4.81  0.21 1.13  0.10 3.09  0.23 1.26 (0.95 ±1.57) 5.10 (3.45 ±17.23)

GHDA baseline

P

Controls

6.82  0.21 1.33  0.12 4.66  0.21 1.70 (1.21 ±2.61) 4.20 (1.68 ±9.70)

< 0.001 0.23 < 0.001 0.004

4.81  0.30 1.22  0.19 3.01  0.34 1.26 (0.86 ±1.63) 4.00 (2.98 ±11.30)

0.26

12 months GH

P

6.03  0.30 1.12  0.09 4.11  0.28 1.64 (1.38 ±2.05) 6.10 (2.24 ±33.15)

0.01 0.62 0.02 0.067 0.28

HDL ˆ high-density lipoprotein; LDL ˆ low-density lipoprotein.

Table 3 Mean  s.e.m. values of serum lipid and lipoproteins in growth hormone (GH) de®cient adults (GHDA) at baseline vs 12 months treatment with GH or placebo. For triglycerides (TG) and lipoprotein (a) (Lp(a)) medians and 25±75 percentiles are given GH Baseline Cholesterol (mmol/L) HDL cholesterol (mmol/L) LDL cholesterol (mmol/L) TG (mmol/L) Lp(a) (mg/dl)

6.83  0.27 1.40  0.21 4.65  0.26 1.38 (1.23 ±2.57) 2.40 (1.68 ±8.05)

Placebo 12 months 6.03  0.30* 1.12  0.09 4.11  0.28 1.64 (1.38 ±2.05) 6.10 (2.24 ±33.15)*

Baseline

12 months

6.80  0.33 1.26  0.14 4.68  0.33 1.82 (0.99 ±2.62) 6.35 (1.68 ±10.0)

6.87  0.33 1.09  0.14 4.89  0.31 1.58 (1.05 ±2.69) 6.50 (4.30 ±18.50)

DGH vs DPlacebo 7 0.79  0.19 7 0.38  0.25 7 0.47  0.31 7 0.05  0.23 10.66  4.09

0.07  0.27* 7 0.17  0.15 0.21  0.29 0.05  0.20 2.10  2.06

*P < 0.05. HDL ˆ high-density lipoprotein; LDL ˆ low-density lipoprotein.

(P ˆ 0.037), but an increase was also seen in the placebo group, although to a lesser extent (Figure 1) so no overall treatment effect was observed (Figure 2). Serum HDL cholesterol and TG were unchanged. No difference was seen in serum IGFBP-1 after GH. Serum insulin increased from 9.1  1.6 mU=l to 15.7  2.6 mU=l after GH and did not change after placebo, with a signi®cant difference in delta values (P ˆ 0.01).

Figure 1 Concentration of lipoprotein (a) (Lp(a)) at baseline and after 12 months of treatment with growth hormone (GH) or placebo, respectively. *P < 0.05.

After 12 months of GH treatment, serum total cholesterol and LDL cholesterol were still signi®cantly higher than in normal subjects (Table 2). In the normal subjects, serum total cholesterol correlated positively with age (r ˆ 0.54, P < 0.001), subcutaneous abdominal fat (r ˆ 0.53, P < 0.001), and intraabdominal fat  mass (r ˆ 0.47, P ˆ 0.005), and inversely with VO2-max (r ˆ 7 0.47, P ˆ 0.005). Positive correlations of borderline signi®cance were found with total body fat (DEXA) (r ˆ 0.40, P ˆ 0.015), and BMI (r ˆ 0.33, P ˆ 0.035). Serum

Figure 2 Delta values of lipoprotein (a) (Lp(a)) after 12 months of treatment with growth hormone (GH) or placebo, respectively.

Lipoproteins in GH-de®cient adults N Vahl et al

LDL cholesterol correlated positively with age (r ˆ 0.45,  P ˆ 0.003). Serum TG correlated negatively with VO2-max (r ˆ 7 0.51, P ˆ 0.002), and positively with intra-abdominal fat mass (r ˆ 0.48, P ˆ 0.003), WHR (r ˆ 0.45, P ˆ 0.003) and age (r ˆ 0.42, P ˆ 0.006). No correlations were found between HDL cholesterol or Lp(a) and either of the variables. No correlations were found between serum TG or lipoproteins and serum IGF-I, IGFBP-1 or insulin. Among the patients at baseline, signi®cant positive correlations were found between serum TG and subcutaneous abdominal fat (r ˆ 0.55, P ˆ 0.003), and serum insulin (r ˆ 0.54, P ˆ 0.004). No correlations were found between either of the lipoproteins and age, body composition, exercise capacity, serum IGF-I or serum insulin. After 12 months no correlations were found between delta HDL cholesterol, delta LDL cholesterol or delta Lp(a) and changes in either body composition, exercise capacity, serum IGF-I or serum insulin. Correlations between delta values of serum cholesterol and serum insulin (r ˆ 0.61, P ˆ 0.037) and between delta values of serum TG and WHR (r ˆ 0.62, P ˆ 0.03) approached signi®cance, whereas no correlations were found between these and either of the other variables.

Discussion This study was designed to compare serum levels of lipid and lipoproteins before and after 12 months of GH treatment in a randomized, double-blind, placebocontrolled trial with values in age- and gendermatched normal subjects, and to see whether changes in the lipid pro®le were related to changes in body composition. Our main ®ndings were that serum levels of cholesterol, LDL cholesterol and TG were signi®cantly increased in GHD adults, and despite a signi®cant decrease in the lipoproteins, a complete normalization after GH treatment was not attained. Serum HDL cholesterol and Lp(a) were within normal ranges and did not change signi®cantly after GH treatment. Moreover, we found no evidence of an association between changes in body composition and lipoproteins. Our ®nding that GHD patients suffer from hypercholesterolaemia caused by an increased concentration of LDL cholesterol is consistent with most,2,9±12,26 but not all,13 studies. The latter study comprised a large group with a wider age range, and it is possible that differences in cholesterol levels between patients and normal subjects are diminished with age. Normal2,10,26 or increased11,13 concentrations of HDL cholesterol, and high9,13,26 or normal2,10,12 levels of serum TG in GHD adults have previously been reported. To date, only one study15 has compared Lp(a) values to normal controls,

reporting values within the normal range in GHD patients, in accordance with our ®ndings. Thus, the results of the effects of GH treatment on circulating lipids in GHD adults are somewhat con¯icting, although there is general agreement of a lowering of total cholesterol. It has been suggested that a reduction in cholesterol only occurs in patients with preexistent hypercholesterolaemia and that the LDL lowering effects are dose dependent.27 All except one patient in our study exhibited a reduction in serum cholesterol without any association to the baseline level, so the effect appears not to be restricted to patients with hypercholesterolaemia. Most of the studies have used quite high doses of GH, but doses between 0.03 IU=kg and 0.07 IU=kg all had a lowering effect on LDL cholesterol. It is, therefore, until now, not con®rmed that the effect of GH on LDL cholesterol is dose dependent. Prospective dose response studies will be necessary to decide this. Our ®nding that HDL cholesterol was unchanged after GH treatment is in accordance with some,11,17±19,28 but not with other,15,16,26,29,30 studies in which increased levels were found. The reason for this discrepancy is not readily explained. All studies, including the present, agree that GH causes no change in serum TG.4,15 ±17,19,28,29,31,32 No overall changes in Lp(a) after GH treatment were found in our study, which is in agreement with some previous studies.17,19 Garry et al18 found no increase in the cohort as a whole, but when dividing patients into those having baseline Lp(a) < 20 mg=dl and > 20 mg=dl, they found a signi®cant increase in Lp(a) in the patients with the higher baseline values. This could not be con®rmed by our data, in which only one subject in each group had Lp(a) values above 20 mg=dl. However, when making the same comparison with Lp(a) values above or below 15 mg=dl a signi®cant increase in the group with high baseline values was demonstrated (P ˆ 0.03, data not shown), whereas no increase was observed in patients with lower baseline values of Lp(a). The plasma level of Lp(a) is mainly genetically determined and inversely related to the apo(a) phenotype and it appears that the in¯uence of other factors are minor.22 We did not measure apo(a) phenotypes. Distribution of Lp(a) is known to be highly skewed with the major part of the population having very low values.33 Forty-eight percent of both the controls and GHD patients in our study had Lp(a) values < 5 mg=dl (Figure 3). In the study by EdeÂn et al16 only two of the nine GHDA patients had Lp(a) values < 5 mg=dl. The larger amount of high baseline values could therefore explain the signi®cant increase in Lp(a) in that particular study. Johannsson et al15 also found an increase in Lp(a) in an open designed study. The necessity of a placebo group to assess the true treatment effect is clearly demonstrated by our results, where an increase was found in the GH treated group, but owing to a small increase in the placebo group no treatment effect was detected. All studies in healthy adults

533

Lipoproteins in GH-de®cient adults N Vahl et al

534

Figure 3. Distribution of lipoprotein (a) (Lp(a)) in the normal controls and in growth hormone de®cient (GHD) patients.

®nd an increase in Lp(a) after treatment with GH.34 ±37 These studies were, however, of very short duration (5 ±14 d), and some used very high doses of GH.36,37 Furthermore, in at least two of the studies34,35 a very high proportion of the subjects had high baseline values of Lp(a). It has recently been reported that IGF-I administration signi®cantly lowers Lp(a) levels,34 which suggests that the increase in Lp(a) seen in GH administration studies are not mediated by IGF-I. It is noteworthy, that insulin levels usually increase with GH administration and become suppressed during IGF-I administration. The importance of studies that evaluate the long-term effect of GH treatment in a maintenance dose as low as possible, is underlined by the fact that Lp(a) is known to be an independent risk factor for atherosclerotic vascular disease,6 even though Ridker et al,38 in a large prospective study, recently found no evidence of such an association. The positive correlations between serum cholesterol and TG, and indices of adiposity, especially intraabdominal fat mass in the normal subjects, con®rm earlier ®ndings in healthy adults.39 ± 41 in a recent study by Snel et al, where magnetic resonance imaging was used to assess body composition, the intraabdominal fat mass correlated positively only with serum TG and not with total cholesterol. This latter study,26 on the other hand, found a negative correlation between intra-abdominal fat and HDL cholesterol, which we could not con®rm. In GHD adults,

very few studies have systematically related lipid status to body composition. We found that baseline levels of serum TG correlated positively with subcutaneous abdominal fat, whereas no correlations were found between baseline lipoproteins and body composition, which con®rms the ®ndings of Cuneo et al,11 where baseline TG levels correlated positively with abdominal skinfold thickness. The changes in lipid and lipoprotein levels, however, were not signi®cantly correlated with changes in adiposity measured by skinfold thickness and total body potassium. In the above mentioned study by Snel et al,26 intraabdominal fat mass in the GHD adults at baseline, correlated positively with serum TG, and negatively with HDL and LDL cholesterol, respectively. In the same study,26 a negative correlation between changes in intra-abdominal fat and in total cholesterol after GH for six months was found. By multiple regression analyses Johannsson et al15 found that changes in body composition (measured by bioelectrical impedance), serum free T4, serum IGF-I and serum insulin, together all explained 24 ±25% of the changes occurring in total cholesterol, LDL cholesterol and Lp(a), whereas changes in the metabolic variables explained 27% and 25% of the changes in apo A1 and TG, respectively. Although correlations (or lack of them) do not imply causality, our study supports and extends the observations that GH exerts speci®c effects on lipoproteins independent of changes in body composition and physical ®tness. GH has been shown to induce hepatic LDL receptor expression in both rats42 and humans,43 and this effect is not mediated by serum IGF-I.4,44 Whether these intriguing data provide an explanation for the direct effect of GH on lipoproteins deserves to be addressed in the future. Acknowledgements

We are grateful to Novo Nordisk, Gentofte, Denmark, for the generous supply of Norditropin. We thank technician Lisbeth Thingholm, Aarhus Kommunehospital for performing the CT-scans, and we are indebted to laboratory technician Eva Seier for her skillful assistance. References

1 RoseÂn T, Bengtsson BA. Premature mortality due to cardiovascular disease in hypopituitarism. Lancet 1990; 336: 285 ± 288. 2 Markussis V, Beshyah SA, Fisher C, Sharp P, Nicolaides AN, Johnston DG. Detection of premature atherosclerosis by highresolution ultrasonography in symptom-free hypopituitary adults. Lancet 1992; 340: 1188 ± 1192. 3 Jùrgensen JOL, Pedersen SA, Thuesen L, Jùrgensen J, Ingemann Hansen T, Skakkebñk NE, Christiansen JS. Bene®cial effects of growth hormone treatment in GH-de®cient adults. Lancet 1989; 1: 1221 ± 1225. 4 Salomon F, Cuneo RC, Hesp R, SoÈnksen PH. The effects of treatment with recombinant human growth hormone on body composition and metabolism in adults with growth hormone de®ciency. N Engl J Med 1989; 321: 1797 ± 1803.

Lipoproteins in GH-de®cient adults N Vahl et al

5 Bengtsson BA, EdeÂn S, LoÈnn L, Kvist H, Stokland A, Lindstedt G, Bosaeus I, Tolli J, SjoÈstroÈm L, Isaksson OG. Treatment of adults with growth hormone (GH) de®ciency with recombinant human GH. J Clin Endocrinol Metab 1993; 76: 309 ± 317. 6 Snel YE, Brummer RJ, Doerga ME, Zelissen PM, Bakker CJ, Hendriks MJ, Koppeschaar HP. Adipose tissue assessed by magnetic resonance imaging in growth hormone-de®cient adults: the effect of growth hormone replacement and a comparison with control subjects. Am J Clin Nutr 1995; 61: 1290 ± 1294. 7 Lapidus L, Bengtsson C, Larsson B, Pennert K, Rybo E, SjoÈstroÈm L. Distribution of adipose tissue and risk of cardiovascular disease and death: a 12 year follow up of participants in the population study of women in Gothenburg, Sweden. BMJ 1984; 289: 1257 ± 1261. 8 Havel RJ. Lowering cholesterol, 1988. Rationale, mechanisms, and means. J Clin Invest 1988; 81: 1653 ± 1660. 9 Merimee TJ, Hollander W, Fineberg SE. Studies of hyperlipidemia in the HGH-de®cient state. Metabolism 1972; 21: 1053 ± 1061. 10 de Boer H, Blok GJ, Voerman HJ, Phillips M, Schouten JA. Serum lipid levels in growth hormone-de®cient men. Metabolism 1994; 43: 199 ± 203. 11 Cuneo RC, Salomon F, Watts GF, Hesp R, SoÈnksen PH. Growth hormone treatment improves serum lipids and lipoproteins in adults with growth hormone de®ciency. Metabolism 1993; 42: 1519 ± 1523. 12 Libber SM, Plotnick LP, Johanson AJ, Blizzard RM, Kwiterovich PO, Migeon CJ. Long-term follow-up of hypopituitary patients treated with human growth hormone. Medicine Baltimore 1990; 69: 46 ± 55. 13 RoseÂn T, EdeÂn S, Larson G, Wilhelmsen L, Bengtsson BA. Cardiovascular risk factors in adult patients with growth hormone de®ciency. Acta Endocrinol Copenh 1993; 129: 195 ± 200. 14 SjoÈberg A, Oscarsson J, Olofsson SO, EdeÂn S. Insulin-like growth factor-I and growth hormone have different effects on serum lipoproteins and secretion of lipoproteins from cultured rat hepatocytes. Endocrinology 1994; 135: 1415 ± 1421. 15 Johannsson G, Oscarsson J, RoseÂn T, Wiklund O, Olsson G, Wilhelmsen L, Bengtsson BA. Effects of 1 year of growth hormone therapy on serum lipoprotein levels in growth hormone-de®cient adults. In¯uence of gender and Apo(a) and ApoE phenotypes. Arterioscler Thromb Vasc Biol 1995; 15: 2142 ± 2150. 16 EdeÂn S, Wiklund O, Oscarsson J, RoseÂn T, Bengtsson BA. Growth hormone treatment of growth hormone-de®cient adults results in a marked increase in Lp(a) and HDL cholesterol concentrations. Arterioscler Thromb 1993; 13: 296 ± 301. 17 Russell Jones DL, Watts GF, Weissberger A, Naoumova R, Myers J, Thompson GR, SoÈnksen PH. The effect of growth hormone replacement on serum lipids, lipoproteins, apolipoproteins and cholesterol precursors in adult growth hormone de®cient patients. Clin Endocrinol Oxf 1994; 41: 345 ± 350. 18 Garry P, Collins P, Devlin JG. An open 36-month study of lipid changes with growth hormone in adults: lipid changes following replacement of growth hormone in adult acquired growth hormone de®ciency. Eur J Endocrinol 1996; 134: 61 ± 66. 19 Modigliani E. RK, B. Uzzan. Modi®cation of blood lipids and lipoproteins after human growth hormone treatment in adults with growth hormone de®ciency: a preliminary report. Endocrinol & Metab 1994; 1(Suppl A): 31 ± 35. 20 Frystyk J, Dinesen B, érskov H. Non-competitive time resolved immuno¯uometric assay for determination of human insulin-like growth factor I and II. Growth Reg 1995; 5: 169 ± 176.

21 Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem 1972; 18: 499 ± 502. 22 Klausen IC, Beisiegel U, Menzel HJ, Rosseneu M, Nicaud V, Fñrgeman O. Apo(a) phenotypes and Lp(a) concentrations in offspring of men with and without myocardial infarction. The EARS Study. European Atherosclerosis Research Study. Arterioscler Thromb Vasc Biol 1995; 15: 1001 ± 1008. 23 Vahl N, Jùrgensen JOL, Jurik AG, Christiansen JS. Abdominal adiposity and physical ®tness are major determinants of the age-associated decline in stimulated GH secretion in healthy adults. J Clin Endocrinol Metab 1996; 81: 2209 ± 2215. Ê strand I. Aerobic work capacity in men and women with 24 A special reference to age. Acta Physiol Scand 1960; 49(Suppl. 169): 9 ± 92. 25 Jùrgensen JOL, Vahl N, Hansen TB, Thuesen L, Hagen C, Christiansen JS. Growth hormone versus placebo treatment for one year in growth hormone de®cient adults: increase in exercise capacity and normalization of body composition. Clin Endocrinol 1996; 45: 681 ± 688. 26 Snel YEM, Doerga ME, Brummer RM, Zelissen PMJ, Koppeschaar HPF. Magnetic Resonance Imaging-Assessed Adipose Tissue and Serum Lipid and Insulin Concentrations in Growth Hormone-De®cient Adults. Arterioscler Thromb Vasc Biol 1995; 15: 1543 ± 1548. 27 de Boer H, Blok GJ, Van der Veen EA. Clinical aspects of growth hormone de®ciency in adults. Endocr Rev 1995; 16: 63 ± 86. 28 Binnerts A, Swart GR, Wilson JH, Hoogerbrugge N, Pols HA, Birkenhager JC, Lamberts SW. The effect of growth hormone administration in growth hormone de®cient adults on bone, protein, carbohydrate and lipid homeostasis, as well as on body composition. Clin Endocrinol Oxf 1992; 37: 79 ± 87. 29 RoseÂn T, Johannsson G, Hallgren P, Caidahl K, Bosaeus I, Bengtsson BA. Bene®cial effects of 12 months replacement therapy with recombinant human growth hormone to growth hormone de®cient adults. Endocrinol Metab 1994; 1: 55 ± 66. 30 Beshyah SA, Henderson A, Niththyananthan R, Skinner E, Anyaoku V, Richmond W, Sharp P, Johnston DG. The effects of short and long-term growth hormone replacement therapy in hypopituitary adults on lipid metabolism and carbo hydrate tolerance. J Clin Endocrinol Metab 1995; 80: 356 ± 363. 31 Whitehead HM, Boreham C, McIlrath EM, Sheridan B, Kennedy L, Atkinson AB, Hadden DR. Growth hormone treatment of adults with growth hormone de®ciency: results of a 13-month placebo controlled cross-over study. Clin Endocrinol Oxf 1992; 36: 45 ± 52. 32 Weaver JU, Monson JP, Noonan K, John WG, Edwards A, Evans KA, Cunningham J. The effect of low dose recombinant human growth hormone replacement on regional fat distribution, insulin sensitivity, and cardiovascular risk factors in hypopituitary adults. J Clin Endocrinol Metab 1995; 80: 153 ± 159. 33 Loscalzo J. Lipoprotein(a). A unique risk factor for atherothrombotic disease. Arteriosclerosis 1990; 10: 672±679. 34 Olivecrona H, Johansson AG, Lindh E, Ljunghall S, Berglund L, Angelin B. Hormonal regulation of serum lipoprotein(a) levels. Contrasting effects of growth hormone and insulin-like growth factor-I. Arterioscler Thromb Vasc Biol 1995; 15: 847 ± 849. 35 Oscarsson J, Ottosson M, Wiklund O, Marin P, Vikman Adolfsson K, BjoÈrntorp P, Eden S. Low dose continuously infused growth hormone results in increased lipoprotein(a) and decreased low density lipoprotein cholesterol concentrations in middle-aged men. Clin Endocrinol Oxf 1994; 41: 109 ± 116. 36 Olivecrona H, Ericsson S, Berglund L, Angelin B. Increased concentrations of serum lipoprotein (a) in response to growth hormone treatment. BMJ 1993; 306: 1726 ± 1727.

535

Lipoproteins in GH-de®cient adults N Vahl et al

536

37 Hansen PS, Kassem M, Brixen K, Klausen IC, Mosekilde L, Fñrgeman O. Effect of short-term treatment with recombinant human growth hormone on lipids and lipoproteins in women and men without growth hormone disturbances. Metabolism 1995; 44: 725 ± 729. 38 Ridker PM, Hennekens CH, Stampfer MJ. A prospective study of lipoprotein(a) and the risk of myocardial infarction. JAMA 1993; 270: 2195 ± 2199. 39 Kissebah AH, Vydelingum N, Murray R, Evans DJ, Hartz AJ, Kalkhoff RK, Adams PW. Relation of body fat distribution to metabolic complications of obesity. J Clin Endocrinol Metab 1982; 54: 254 ± 260. 40 Katzel LI, Busby Whitehead MJ, Goldberg AP. Adverse effects of abdominal obesity on lipoprotein lipids in healthy older men. Exp Gerontol 1993; 28: 411 ± 420.

41 Pedersen SB, BoÈrglum JD, Brixen K, Richelsen B. Relationship between sex hormones, body composition and metabolic risk parameters in premenopausal women. Eur J Endocrinol 1995; 133: 200 ± 206. 42 Rudling M, Norstedt G, Olivecrona H, Reihner E, Gustafsson JA, Angelin B. Importance of growth hormone for the induction of hepatic low density lipoprotein receptors. Proc Natl Acad Sci USA 1992; 89: 6983 ± 6987. 43 Angelin B, Rudling M, Olivecrona H, Ericsson S. Effects of growth hormone on low-density lipoprotein metabolism. Acta Paediatr Suppl 1992; 383: 67 ± 68. 44 Rudling M, Olivecrona H, Eggertsen G, Angelin B. Regulation of rat hepatic low density lipoprotein receptors. In vivo stimulation by growth hormone is not mediated by insulin-like growth factor I. J Clin Invest 1996; 97: 292 ± 299.