HAL Id: hal-02673785
https://hal.inrae.fr/hal-02673785
Submitted on 31 May 2020
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Impact of growth hormone (GH) treatment on cardiovascular risk factors in GH-deficient adults: a metaanalysis of blinded, randomized, placebo-controlled
trials
P Maison, S Griffin, M Nicoue-Beglah, Nadia Haddad, B Balkau, P Chanson
To cite this version:
P Maison, S Griffin, M Nicoue-Beglah, Nadia Haddad, B Balkau, et al.. Impact of growth hormone
(GH) treatment on cardiovascular risk factors in GH-deficient adults: a metaanalysis of blinded,
randomized, placebo-controlled trials. Journal of Clinical Endocrinology and Metabolism, Endocrine
Society, 2004, 89 (5), pp.2192-2199. �hal-02673785�
Impact of Growth Hormone (GH) Treatment on
Cardiovascular Risk Factors in GH-Deficient Adults:
A Metaanalysis of Blinded, Randomized, Placebo- Controlled Trials
PATRICK MAISON, SIMON GRIFFIN, MARC NICOUE-BEGLAH, NABILA HADDAD, BEVERLEY BALKAU, AND PHILIPPE CHANSON
Clinical Pharmacology (P.M.), Clinical Research Unit, Assistance Pubique-Hoˆpitaux de Paris, Henri Mondor University Hospital, F-94010 Cre´teil, France; Institut National de la Sante´ et de la Recherche Me´dicale U258 (P.M., M.N.-B., N.H., B.B.), F-94807 Villejuif, France; Institute of Public Health (S.G.), CB2 2SR Cambridge, United Kingdom; and
Endocrinology and Reproductive Diseases (P.C.), Assistance Publique-Hoˆpitaux de Paris, Biceˆtre University Hospital, and University Paris XI, F-94275 Le Kremlin-Biceˆtre, France
Patients with hypopituitarism have an increased risk of cardio- vascular mortality. GH treatment could modify the cardiovas- cular risk in adults with GH deficiency, but most published clin- ical trials involved few patients and the results are variable.
We conducted a systematic review of blinded, randomized, placebo-controlled trials of GH treatment in adult patients with GH deficiency published up to August 2003. Thirty-seven trials were identified. We combined the results for effects on lean and fat body mass; body mass index; triglyceride and cholesterol [high-density lipoprotein, low-density lipoprotein (LDL), and total] levels; blood pressure; glycemia; and insu- linemia. Overall effect size was used to evaluate significance, and weighted differences between GH and placebo were used to appreciate the size of the effect.
GH treatment significantly reduced LDL cholesterol [ ⴚ 0.5 (
SD0.3) mmol/liter], total cholesterol [ ⴚ 0.3 (0.3) mmol/liter], fat mass [ ⴚ 3.1 (3.3) kg], and diastolic blood pressure [ ⴚ 1.8 (3.8) mm Hg] and significantly increased lean body mass [ ⴙ 2.7 (2.6) kg], fasting plasma glucose [ ⴙ 0.2 (0.1) mmol/liter], and insulin [ ⴙ 8.7 (7.0) pmol/liter]. All effect sizes remained significant in trials with low doses and long-duration GH treatment.
Thus, GH treatment has beneficial effects on lean and fat body mass, total and LDL cholesterol levels, and diastolic blood pressure but reduces insulin sensitivity. The global car- diovascular benefit remains to be determined in large trials with appropriate clinical endpoints. (J Clin Endocrinol Metab 89: 2192–2199, 2004)
S INCE THE ADVENT of recombinant GH in 1985, GH therapy has been extended to adults with pituitary de- ficiency, in whom GH deficiency (GHD) has been hypoth- esized to be a cardiovascular risk factor (1– 4). An adult GHD syndrome was recently described (5), in which patients ex- hibit cardiovascular risk factors such as abdominal obesity, hypercholesterolemia, and hypertriglyceridemia. GH re- placement therapy may reduce some of these cardiovascular risk factors (reviewed in Ref. 6). Adverse effects include insulin resistance and increased volemia (7). Because hyper- tension and diabetes are common in patients with acromeg- aly and are also well-known cardiovascular risk factors, these effects are of particular interest. Most clinical trials of GH therapy in GH-deficient adults have involved small numbers of patients. Reported effects on cardiovascular risk factors varied according to the dose and duration of GH therapy, age at onset of GHD, and gender.
To obtain a more reliable picture of the effects of GH treatment on the main cardiovascular risk factors in GH- deficient adults (body mass, lipids, blood pressure, plasma
glucose, and insulin), we conducted a systematic review of all blinded, randomized, placebo-controlled trials of GH in adults with GHD published up to August 2003.
Subjects and Methods Identification of relevant trials
We searched the Medline (Ovid), Experta Medica (EMBASE), and Biosis electronic databases, from their year of inception to August 2003.
The medical literature was searched for all reports containing the key words, growth hormone (or somatotropin), trial, and human. The search strategy was not limited by study design or language.
A manual search of the Journal of Clinical Endocrinology and Metabolism since 1985 was used to assess the sensitivity of our electronic search.
Additional information was requested from GH manufacturers, refer- ences cited in published articles, and clinical trials investigators.
Inclusion criteria
We included all randomized, blinded, placebo-controlled trials in- volving patients aged over 17 yr with GHD corresponding to less than 5 g/liter after stimulation, as recommended by the consensus guide- lines for the diagnosis and treatment of adults with GHD (8).
Included trials had at least one of the following outcome measures:
diastolic blood pressure, systolic blood pressure, fasting blood glucose, fasting insulinemia, triglyceridemia, cholesterolemia [high-density li- poprotein, low-density lipoprotein (LDL), or total], lean body mass, fat mass (when expressed in absolute terms), and body mass index (BMI).
These outcomes are the main cardiovascular risk factors in this setting and are the most frequently reported in clinical trials. Lean body mass Abbreviations: BMI, Body mass index; GHD, GH deficiency; LDL,
low-density lipoprotein.
JCEM is published monthly by The Endocrine Society (http://www.
endo-society.org), the foremost professional society serving the en- docrine community.
doi: 10.1210/jc.2003-030840
2192
is not, strictly speaking, a direct cardiovascular risk factor, but, because of the relationship between BMI, fat mass, and lean body mass, the effects of GH on these three parameters are of interest. From more than 3000 published reports, one of the authors (P.M.) selected the abstracts of trials that potentially met the inclusion criteria. The corresponding articles were then checked for the inclusion criteria, independently by three authors (N.H., M.N.-B., and P.M.). Discrepancies were resolved through discussion with all the authors. To detect a possible publication bias, we asked GH manufacturers what proportion of randomized, placebo-controlled trials were finally published.
Data extraction and outcomes
Three metaanalysts (N.H., M.N.-B., and P.M.) extracted data from published reports to a standard form. Authors were contacted to verify the extracted data where necessary. Discrepancies were resolved by discussion among all the authors of the present paper. The following data were extracted: population characteristics (setting, age, sex, num- ber, weight, BMI, and disease onset), treatment (dose, duration, and frequency), study quality (design, randomization method, blinding, pla- cebo vials, and statistical methods), losses to follow-up (for each out- come measure), baseline and follow-up values and changes (means and sd or sem), and methods used to measure outcomes. All data extracted were summary data.
Statistical methods
For primary analyses of continuous outcome measures, we first cal- culated standardized effect sizes for each trial and then the global effect size for each outcome (9). The effect size is a measure of the overlap in the distribution of outcome scores between two treatment groups. The effect size was calculated differently for parallel-group and cross-over studies, to reflect intergroup and intragroup comparisons (9). For par- allel groups, the effect size was computed as the mean difference (GH minus placebo) in the changes (follow-up minus baseline) for each outcome divided by the estimated variance of changes in the two groups.
For cross-over trials, the effect size was calculated as the mean difference in values at the end of each period divided by the variance in the placebo group at follow-up. In some instances, these values had to be estimated from graphs in the articles [four for lean body mass (10 –14), two for insulin (10, 11), one for glucose (10), and three for blood pressure (15–17).
To calculate the overall effect size, the effect size in each study was weighted by the reciprocal of the variance. We present these scores with their 95% confidence intervals. A positive effect size implies an increase in the frequency of the outcome with GH treatment, and a negative effect size implies a decrease.
Because the variances for changes were not directly reported in all articles, they were calculated from t statistics, probability values, or confidence intervals (variances) for the GH and placebo groups (parallel design) or the study period (cross-over design) (9). We used a Q test to explore heterogeneity between studies. The analyses were repeated us- ing a random-effects model when the effect size was significant in a fixed model (18). Because the random-effects model incorporates statistical heterogeneity (results, methodology, and publication bias) and provides a more conservative estimate of the pooled effect size than a fixed model, we present all the results of effect-size according to a random model.
Funnel plots were drawn and their asymmetry was assessed to deter- mine the possible influence of publication and location biases (19). The intercept of the weighted and unweighted linear regression lines, when the effect size divided by the se is regressed against the reciprocal of the se, provides a measure of asymmetry. Because the effect size may be significant because of a single trial (e.g. large trials or trials with large effects), we also conducted a sensitivity analysis. When the effect was carried by one or two trials, these studies were dropped from the anal- ysis to verify whether the same trend was observed with the remaining trials. To quantify the size of the effect, we present the weighted (by the variance) mean difference (and sd) between the GH and placebo groups for each outcome measure.
The effects of the GH dose, GH treatment duration, percentage of patients with adult onset, and study design on overall estimates were assessed by stratification or metaregression. Weighted least-squares re- gression analysis was used for metaregression, individual study effects being weighted by the reciprocal of the estimated variance. The -
coefficient and its significance are presented, along with the adjusted R
2value, to show the overall variability explained by the model. Analyses were conducted using the SPSS (SPSS Inc., Chicago, IL) for Windows package.
Results
The combined search strategy identified 37 blinded, ran- domized, placebo-controlled trials of GH in GH-deficient adults that included at least one of the necessary outcome measures (Tables 1 and 2) (7, 10 –17, 20 –54). Different pub- lications describing the same trial may refer to different out- comes. Eight trials could not be included because the re- quired summary data could not be obtained. One trial was single blinded (13, 14). The trials were generally of good quality. Median losses to follow-up were 10% of patients [95% confidence interval ⫽ (0, 57)], and data were rarely analyzed on an intention-to-treat basis: their low number did not allow pertinent specific analysis of this subgroup of trials.
Most reported comparisons were pretreatment vs. posttreat- ment rather than GH vs. placebo. Whatever the outcome measure, no significant heterogeneity was observed (Table 3). However, to obtain more conservative estimates, we present effect sizes with random-effects models. Funnel plot- ting and linear regression suggested a selection bias for in- sulin only (intercept, 9.4; P ⫽ 0.03 with weighted regression, and intercept, 7.3; P ⫽ 0.02 with unweighted regression), but publication bias was unlikely when this outcome was re- ported because it rarely reached statistical significance in individual trials. Three of the five GH manufacturers in- formed us that all the randomized, placebo-controlled trials they sponsored were finally published; the other two man- ufacturers did not answer or were unable to provide us with this information.
Body mass
A significant positive overall effect on lean body mass was found [0.45 (0.32; 0.58)]. The weighted mean difference in lean body mass was ⫹ 2.74 (2.61) kg (Table 3). The overall effect size was significant in trials (n ⫽ 6) based on
40K counting [0.50 (0.05; 0.94)], trials (n ⫽ 7) based on bioelectrical impedance [0.60 (0.40; 0.80)], and trials (n ⫽ 6) based on dual-energy x-ray absorptiometry [0.46 (0.22; 0.71)]. Five of the 19 trial reports that included lean body mass also in- cluded changes in total body mass. In these studies, the overall effect size between the GH and placebo groups was not significant for total body mass [ ⫺ 0.07 ( ⫺ 0.31; 0.17)] and remained similar for lean body mass [0.46 (0.30; 0.61)].
The overall effect size was significantly negative for fat mass [ ⫺ 0.62 ( ⫺ 0.78; ⫺ 0.48)], with a weighted mean dif- ference of ⫺ 3.05 kg (3.29). No effect was found on BMI (Table 3).
Lipids
Significant beneficial effects were observed only for LDL and total cholesterol (Table 3). Weighted mean differences were ⫺ 0.53 (0.29) mmol/liter and ⫺ 0.34 (0.31) mmol/liter, respectively. Effect sizes were ⫺ 0.35 ( ⫺ 0.52; ⫺ 0.17) and
⫺ 0.24 ( ⫺ 0.39; ⫺ 0.08), respectively. In the analysis of sensi- tivity for LDL cholesterol, excluding the two trials (27, 48)
Maison et al. • GH and Cardiovascular Risk Factors J Clin Endocrinol Metab, May 2004, 89(5):2192–2199 2193
with the most marked effects on this outcome measure, the overall effect size was still significant [ ⫺ 0.26 ( ⫺ 0.47; ⫺ 0.05)].
Blood pressure
Ten trials involving a total of 401 patients were included in the metaanalysis (Table 3). Ambulatory blood pressure was measured in two trials (15, 41). In the other eight trials, blood pressure was measured manually, usually after the participants had been lying or sitting for at least five min. The effect size was not significant for systolic blood pressure. In contrast, the overall effect size was significantly negative for diastolic blood pressure [ ⫺ 0.25 ( ⫺ 0.43; ⫺ 0.07)]. The weighted mean difference in diastolic blood pressure was
⫺ 1.80 (3.77) mm Hg. The effect size was still significant after exclusion of one large trial (27).
Glucose/insulin
With 13 trials involving 511 patients, a significant overall effect of GH treatment on fasting glucose was observed [ ⫹ 0.43 (0.26; 0.60)] (Table 3). The mean weighted difference in blood glucose between the groups was ⫹ 0.22 (0.14) mmol/
liter, but weighted mean fasting glucose values were within the normal range on both GH treatment [5.1 mmol/liter (0.5)]
and placebo [4.8 mmol/liter (0.4)]. In the analysis of sensi- tivity, the effect size was still significant after excluding three large trials (7, 14).
With 11 trials involving 378 patients, a significant overall effect of GH treatment on fasting insulin was observed [ ⫹ 0.42 (0.23; 0.61)] (Table 3). The mean weighted difference in plasma insulin was 8.7 (7.0) pmol/liter between GH and placebo. This difference remained significant after exclusion of one large trial (7).
Effects of the GH dose and treatment duration, gender, age, age at GHD onset, and trial designs
Effect of the GH dose (Table 4). No significant relationship was observed between the GH dose and the effect size in meta- regression analysis. This result could be explained by a nar- row dose distribution. A subgroup analysis was then done with trials using target doses of no more than 0.35 U/kg 䡠 wk and a target dose of 0.5 U/kg 䡠 wk. All effect sizes remained significant in the analysis of low-dose trials (Table 4). A smaller number of trials used high doses, and the effect size in this subgroup was significant only for fat mass, glucose, and insulin. A dose-dependent effect of GH was found on fat TABLE 1. Characteristics of blind, randomized, placebo-controlled trials included in the meta-analyses
First author (Ref.) Year Study design Blind Patients included No. lost
Attanasio, Chipman (7, 20) 1997 Parallel DB 74 16
Attanasio, Chipman (7, 20) 1997 Parallel DB 99 7
aBaum, Sesmilo (13, 14) 1996 Parallel SB 40 5
Bell (21) 1998 Parallel DB 24 4
Bell (21) 1998 Parallel DB 27 4
Beshyah (11, 12) 1995 Parallel DB 40 2
Boger (15) 1996 Parallel DB 30 3
Bramnert (22) 2003 Parallel DB 19 0
Caidhal (23) 1994 Cross-over DB 10 1
Christ (24, 25) 1997 Parallel DB 14 1
aChrist (26) 1999 Parallel DB 21 3
Cuneo (27) 1998 Parallel DB 166 94
aEzzat (28) 2002 Parallel DB 115 12
aFernholm (29) 2000 Parallel DB 31 0
Florkowski (30) 1996 Cross-over DB 20 1
Fowelin (31) 1993 Cross-over DB 9 2
Hoffman (16) 1996 Cross-over DB 8 1
Hwu (32) 1997 Parallel DB 21 5
Johannsson (33, 34) 1996 Parallel DB 68 3
Johansson (35) 2002 Cross-over DB 10 0
Jorgensen (36, 37) 1989 Cross-over DB 22 1
Kousta (38) 1998 Parallel DB 13 0
Leese (39) 1998 Parallel DB 32 2
Mesa (40) 2003 Parallel DB 165
Moller (41) 1999 Parallel DB 24 2
Nass (42) 1995 Parallel DB 20 0
Nolte (43) 1997 Parallel DB 38 6
Rosenfalck (44) 1999 Parallel DB 24
Russell-Jones (45, 46) 1993 Parallel DB 18
Russell-Jones (47) 1998 Parallel DB 12 0
Salomon, Cuneo (10, 48) 1989 Parallel DB 24 5
aSmith (49) 2002 Parallel DB 32 0
Snel (50) 1995 Parallel DB 38 13
Vahl (51) 1998 Parallel DB 27
Valcavi (52) 1995 Parallel DB 10 0
Webster (53) 1997 Parallel DB 18 3
Whitehead (54) 1992 Cross-over DB 14 2
DB, Double blind; SB, single blind.
a
Maximum lost (varying with outcome).
TABLE 2. Patient characteristics, treatment, and outcomes by trials First author (Ref.) Hypopituitarism (%) Adult-onset (%) Women (%) Age (yr)
aTarget dose (U/kg
䡠wk) Duration (months) Outcomes Attanasio, Chipman (7, 20) 0 2 6 28.8 (8.0) 0.25 6 LBM, FM, HDL, LDL, Chol, Ins, Gly Attanasio, Chipman (7, 20) 100 38 43.5 (10.0) 0.25 6 LBM, FM, HDL, LDL, Chol, Ins, Gly Baum (13) 97 100 0 50.0 (8.7) 0.21 18 LBM Baum, Sesmilo (13, 14) 95 100 0 (24 – 64) 0.21 18 HDL, LDL, Chol, TG, Ins, Gly Bell (21) 100 44.4 (8.8) 0.25 6 LBM Bell (21) 0 43.9 (7.2) 0.25 6 LBM Beshyah (11, 12) 100 80 53 (19 – 67) 0.35 6 LBM, FM, HDL, LDL, Chol, TG, BP, G Boger (15) 100 100 53 41.5 (11.0) 0.25 12 HDL, LDL, Chol, TG, BP Bramnert (22) 100 79 37 42.0 (10.9) 0.2 6 BMI, Ins, Gly Caidhal (23) 100 100 10 (34 –58) 0.5 6 B P Christ (24, 25) 100 86 57 47.4 (15.1) 0.25 3 LBM, FM, Ins Christ (26) 100 90 61 48.6 (15.3) 0.25 3 HDL, LDL, Chol, TG Cuneo (27) 100 44 40.5 (13.5) 0.25 6 LBM, HDL, LDL, Chol, TG, BP, Gly Ezzat (28) 42 (20 –70) 0.21 6 LBM, FM Fernholm (29) 100 100 19 (19 –79) 0.1 6 LBM, FM, Ins Florkowski (30) 80 (20 – 69) 0.25 3 BMI Fowelin (31) 100 100 11 48.6 (9.2) 0.5 6 Ins, Gly Hoffman (16) 100 87 13 49.4 (23.4) 0.56 1 w k B P Hwu (32) 95 100 52 29.5 (7.4) 0.25 6 HDL, LDL, Chol, TG, BMI, FM Johannsson (33, 34) 96 69 35 44.3 (9.9) 0.25 6 LBM, BMI, FM Johansson (35) 100 100 10 (48 – 69) 0.21 1 w k B P Jorgensen (36, 37) 50 0 3 6 23.8 (5.6) 0.5 4 BMI, BP, Ins Kousta (38) 100 82 62 46.5 (9.5) 0.25 3 Gly Leese (39) 77 66 35.3 (12.1) 0.25 6 HDL, LDL, Chol, TG Mesa (40) 100 63 40 (19 – 60) 0.25 6 LBM, FM Moller (41) 92 20 36.9 (10.6) 0.5 4 B P Nass (42) 95 100 20 (27– 60) 0.25 6 LBM Nolte (43) 100 100 47 41.6 (10.7) 0.5 12 HDL, LDL, Chol, TG Rosenfalck (44) 95 71 25 38.0 (10.7) 0.5 4 LBM, FM, Ins, Gly Russell-Jones (45, 46) 100 56 46.6 (9.1) 0.25 2 LBM, HDL, LDL, Chol, TG, Ins Russell-Jones (47) 100 100 66 46.8 (6.5) 0.25 2 LBM, BMI, Gly Salomon, Cuneo (10, 48) 33 38.0 (9.8) 0.5 6 LBM, FM, TG, HDL, LDL, Chol, BP, Ins Smith (49) 100 41 44.0 (14.0) 0.14 6 B P Snel (50) 100 100 34 (20 – 60) 0.25 6 LBM, BMI, FM Vahl (51) 74 100 33 44.5 (9.4) 0.5 12 HDL, LDL, Chol, TG Valcavi (52) 100 100 30 47.2 (11.6) 0.35 12 BP Webster (53) 94 33 44.3 (10.0) 0.25 6 Chol, BMI, Gly Whitehead (54) 79 43 36 29.4 (10.1) 0.5 6 Chol, LBM, FM, Gly Ins, Insulinemia; Gly, glycemia; LBM, lean body mass; FM, fat mass; HDL, HDL cholesterol; LDL, LDL cholesterol; Chol, cholesterol; TG, triglyceride s; BP, blood pressure.
aAge, mean (
SD) o r range.
Maison et al. • GH and Cardiovascular Risk Factors J Clin Endocrinol Metab, May 2004, 89(5):2192–2199 2195
mass, with a lower effect size with low-dose GH [ ⫺ 0.2 ( ⫺ 0.4;
⫺ 0.0)] than with high-dose GH [ ⫺ 0.6 ( ⫺ 1.0; ⫺ 0.1)].
Effect of the treatment duration (Table 4). No significant rela- tionship was observed between the treatment duration and the effect sizes in metaregression analysis. Thus, Table 4 shows only the results of subgroup analyses. All effect sizes remained significant in the analysis of trials with long- duration treatment. There were few short-term trials, ex- plaining the smaller number of significant parameters. Over- all, the results pointed to a greater effect of prolonged treatment on lean mass and diastolic blood pressure.
Effect of gender. When comparing effects according to gender, a significant negative relationship was found only between proportion of women and the effect size for blood glucose (  ⫽ ⫺ 0.59, P ⫽ 0.03, adjusted R
2⫽ 0.30), suggesting a lesser effect in women than in men.
Effect of age. A significant relationship was observed between mean age and the effect size for LDL cholesterol (  ⫽ ⫺ 0.60, P ⫽ 0.03, R
2⫽ 0.36) and total cholesterol (  ⫽ ⫺ 0.64, P ⬍ 0.04, R
2⫽ 0.41): the younger the patient, the stronger the effect.
Effect of age at GHD onset. The relationship between the pro- portion of patients with adult-onset (vs. childhood-onset) GHD and the effect of GH treatment was significant for diastolic blood pressure (  ⫽ ⫺ 0.88, P ⫽ 0.02, R
2⫽ 0.71) and plasma insulin (  ⫽ ⫺ 0.76, P ⫽ 0.03, R
2⫽ 0.50), suggesting a greater beneficial effect on blood pressure and a lesser negative impact on insulin in patients with adult-onset GHD.
Effect of the trial design. The overall effect size in subgroup analyses of parallel-group studies remained similar for all outcomes. There were few cross-over studies (one for cho- lesterol and lean body mass, two for insulin and glucose, and four for blood pressure), and the global effect sizes were not significant.
Discussion
This systematic review of 37 blinded, randomized, place- bo-controlled trials shows a small but statistically significant beneficial effect of GH treatment on lean and fat body mass, LDL and total cholesterol, and diastolic blood pressure in GH-deficient adults. In contrast, GH therapy significantly increased plasma glucose and insulin levels.
A major potential source of bias in systematic reviews is that trials with positive results are more likely to be pub- lished than trials with neutral or negative results. In our metaanalysis, this bias seems unlikely with regard to blood pressure and total and LDL cholesterol (rare significant re- sults) and blood glucose and insulin (negative results). Fur- thermore, information provided to us by GH manufacturers suggests that most GH trials in this setting are published.
Another limitation of metaanalyses is the variable quality of the selected trials. To minimize this problem, we selected only studies with protocols meeting strict methodological quality criteria. This, of course, does not rule out quality problems arising during the trials’ progress.
Lean body mass increased and fat mass decreased after GH treatment, whereas total body weight remained constant.
TABLE 3. Results of meta-analysis of GH effects on cardiovascular risk factors
Lean B mass, Lean body mass; TG, triglycerides; Chol., cholesterol; D.B.P., diastolic blood pressure; S.B.P., systolic blood pressure; ns, nonsignificant.
TABLE 4. Effect size (95% confidence interval) in subgroup analysis with trials using target doses of no more than 0.35 U/kg
䡠wk, and with trials with a target dose of 0.5 U/kg
䡠wk and with trials of short duration ( ⬍ 6 months) and with trials of longer duration ( ⱖ 6 months)
Low dose
( ⱕ 0.35 U/kg BW) n High dose
( ⬎ 0.5 U/kg BW) n Short duration
( ⬍ 6 months) n Long duration
( ⱖ 6 months) n Lean body mass 0.5 (0.4; 0.7)
a16 0.2 ( ⫺ 0.2; 0.6) 3 0.2 ( ⫺ 0.5; 0.9) 4 0.6 (0.4; 0.7)
a15 Fat mass ⫺ 0.2 ( ⫺ 0.4; ⫺ 0.0)
a10 ⫺ 0.6 ( ⫺ 1.0; ⫺ 0.1)
a3 ⫺ 0.3 ( ⫺ 0.9; 0.4) 2 ⫺ 0.7 ( ⫺ 0.5; ⫺ 0.8)
a11 Diastolic BP ⫺ 0.3 ( ⫺ 0.6; ⫺ 0.1)
a6 ⫺ 0.1 ( ⫺ 0.4; 0.2) 4 0.1 ( ⫺ 0.4; 0.2) 4 ⫺ 0.4 ( ⫺ 0.6; ⫺ 0.1)
a6 LDL cholesterol ⫺ 0.3 ( ⫺ 0.5; ⫺ 0.1)
a10 ⫺ 0.4 ( ⫺ 0.8; 0.0) 3 ⫺ 0.4 ( ⫺ 0.5; ⫺ 0.1)
a9 ⫺ 0.4 ( ⫺ 0.8; ⫺ 0.1)
a4 Total cholesterol ⫺ 0.2 ( ⫺ 0.4; ⫺ 0.1)
a10 ⫺ 0.3 ( ⫺ 0.6; 0.0) 5 ⫺ 0.2 ( ⫺ 0.3; 0.0)
a11 ⫺ 0.5 ( ⫺ 0.8; ⫺ 0.1)
a4
Glucose 0.4 (0.2; 0.7)
a7 0.5 (0.3; 0.8)
a6 0.5 ( ⫺ 0.1; 1.1) 3 0.5 (0.3; 0.7)
a10
Insulin 0.5 (0.2; 0.7)
a6 0.3 (0.0; 0.6)
a5 0.4 (0.1; 0.7)
a4 0.4 (0.2; 0.7)
a7
n, Number of studies for each subgroup; BP, blood pressure; BW, body weight.
a