- Email: [email protected]
Growth hormone treatment in children with chronic renal failure: A meta-analysis of randomized controlled trials
Growth hormone treatment in children with chronic renal failure: A meta-analysis of randomized controlled trials Dushyanthi Vimalachandra, BA, MPH, Jonathan C. Craig, MBChB, DCH, MMed (Clin Epi), FRACP, PhD, Chris T. Cowell, MBBS, FRCP(C), FRACP, and John F. Knight, MBBS, MA, MBA, FRACP Objective: To evaluate the benefits and side effects of recombinant human growth hormone (hGH) treatment in children with chronic renal failure. Methods: Two reviewers independently assessed relevant randomized controlled trials for methodologic quality, extracted data, and estimated summary treatment effects by use of a random effects model. Results: Ten randomized controlled trials involving 481 children were identified. Treatment with hGH (28 IU/m2/wk) resulted in a significant increase in height standard deviation score at 1 year (4 trials, weighted mean difference [WMD] = 0.77, 95% CI = 0.51 to 1.04), and a significant increase in height velocity at 6 months (2 trials, WMD = 5.7 cm/y, 95% CI 4.4 to 7.0) and 1 year (2 trials, WMD = 4.1 cm/y, 95% CI 2.6 to 5.6), but there was no further increase in height indexes during the second year of administration. Compared with the 14 IU/m2/wk group, there was an increase of 1.4 cm/y (0.6 to 2.2) in height velocity in the group treated with 28 IU/m2/wk. The frequency of reported side effects of hGH were similar to that of the control group. Conclusion: On average, 1 year of treatment with 28 IU/m2/wk hGH in children with chronic renal failure results in an increase of 4 cm/y in height velocity above that of untreated control subjects, but there was no demonstrable benefit for longer courses or higher doses of treatment. (J Pediatr 2001;139:560-7)
As defined by the need for dialysis or kidney transplantation, 32 per 1 million children younger than 15 years have chronic renal failure (CRF)1; many
more children have less advanced CRF (glomerular filtration rate [GFR] < 30 mL/min/1.73 m2). Growth retardation, one of the complications of CRF, is of
From the Centre for Kidney Research, The Children’s Hospital at Westmead, Robert Vines Growth Research Centre, The Children’s Hospital at Westmead, and the Department of Public Health and Community Medicine, University of Sydney, Australia.
Supported by the Australian Kidney Foundation, the Small Grants Scheme of the Royal Alexandra Hospital for Children, and the Cochrane Child Health Field Bursary. Submitted for publication Aug 8, 2000; revisions received Dec 21, 2000, and Feb 20, 2001; accepted May 25, 2001. Reprint requests: Dushyanthi Vimalachandra, Centre for Kidney Research, The Children’s Hospital at Westmead, Locked Bag 4001, Westmead, NSW 2145, Australia. Copyright © 2001 by Mosby, Inc. 0022-3476/2001/$35.00 + 0 9/21/117582 doi:10.1067/mpd.2001.117582
560
concern to the families of more than 90% of children with CRF.2 Approximately 60% of boys and 41% of girls who started renal replacement therapy before age 15 years attain a final adult height more than 2 SD (12 cm) below the mean for healthy adults.3 Impairment of growth can begin when the GFR falls to 50% of normal and becomes an increasing problem once the GFR falls below 25%.4 Over the past 10 years, recombinant CRF GFR hGH MD SDS WMD
Chronic renal failure Glomerular filtration rate Human growth hormone Mean difference Standard deviation score Weighted mean difference
human growth hormone (hGH) treatment has been used to help short children with CRF attain a height more in keeping with their age group.5 However, there are concerns that hGH may have an adverse effect on the preservation of native renal function, predispose to acute rejection in kidney transplant recipients, and cause benign intracranial hypertension and slipped capital femoral epiphysis.6 Although many trials of hGH treatment in children with CRF have been undertaken, uncertainty exists on the magnitude of benefits and side effects of the treatment.5 A systematic review of hGH treatment was undertaken to evaluate growth outcomes to establish the effect of treatment over time. We sought to establish whether the growth outcomes remained linear over time or whether there was a waning effect of
VIMALACHANDRA ET AL
THE JOURNAL OF PEDIATRICS
VOLUME 139, NUMBER 4 the treatment. Second, we examined the effect of varying doses of the treatment. Third, we explored the effect of the following factors on treatment: age, sex, pubertal status, and the stage of CRF (before dialysis, on dialysis, after transplantation). Finally, the study evaluated potential side effects of hGH treatment.
METHODS Inclusion Criteria Randomized controlled trials that compared hGH treatment with placebo or no treatment or that compared 2 doses of hGH treatment were identified. The study participants were under age 18 years, diagnosed with renal failure and either had not yet started dialysis, were on dialysis, or had undergone transplantation. Growth outcomes were assessed by standard deviation score (SDS). An individual’s height SDS was calculated as follows: (Actual height – Mean height for age)/SD at that age. Normal children have a height that ranges between –2 and +2 SDS. If a change in height SDS is negative, it implies growth failure and if positive, catch-up growth. The primary outcome measure was the difference in mean change in height SDS between the treatment and control groups during the trial period. This was computed as [(postmean height SDS(treated) – pre-mean height SDS (treated) ) – (post-mean height SDS(control) – pre-mean height SDS (control) )]. Secondary outcomes included change in height SDS from treatment onset to completion, change in height SDS during puberty, change in height velocity, change in height velocity SDS, final height, change in bone age, quality of life, and adverse effects of hGH treatment (hypertension, graft function, rejection episodes, and any other possible side effects reported).
Search Strategy A systematic and comprehensive literature search was carried out to identify eligible randomized controlled trials.
The Cochrane Controlled Trials Register (Issue 2, 2000) was searched with the following search terms used as medical subject heading and text words: growth hormone/somatotropin, chronic renal failure, end stage renal disease, and kidney transplantation. Two reviewers (D.V. and J.K.) independently searched Medline (1966 to April 2000) and Embase (1988 to April 2000) without language restriction. An optimally sensitive search strategy developed for identification of randomized controlled trials7,8 was combined with medical subject heading and text words for CRF and growth hormone. Additional studies were located through article reference lists and through contact with local and international experts in the field. Two reviewers (D.V. and J.K.) independently screened titles and abstracts of studies identified through the searches and selected trials that met the inclusion criteria. This process favored overselection to avoid losing relevant studies. In the event of uncertainty or no abstract available, the full article was obtained. Any disagreement on study selection was resolved through discussion and consultation with a third reviewer (J.C.).
Data Extraction and Quality Assessment Two reviewers (D.V. and J.K.) independently extracted data from eligible full articles. Information was collected on participant characteristics (number, age, sex, pubertal stage, stage of CRF, comorbidities), intervention (type of growth hormone, dose, duration, cointerventions), and primary and secondary outcome measures. Methodologic quality (allocation concealment, blinding of outcome assessment, completeness of follow-up, and analysis by intention to treat9) of the eligible studies were also assessed. In the case of cross-over trials, data from the first half of the study were extracted and used in the analysis. Authors were contacted to obtain any missing data, as well as raw data when necessary. Any discrepancies in data extraction were discussed with a third re-
viewer (J.C.). In the case where the results of a study were published more than once or the results were discussed in a number of publications, the most complete data were obtained by use of all sources and included in the analysis only once.
Statistical Analysis Difference in means was calculated for continuous outcomes of individual trials to estimate the size of the intervention effects. In the case where a trial reported only 6-month velocity data, the mean and SD were doubled to arrive at annualized data.10 Overall treatment effects (weighted mean difference) were computed by use of a random effects model that takes into account betweenstudy variability, as well as within-study variability. Heterogeneity was analyzed with Cochran’s Q statistic.11 Subgroup analysis was attempted to explore the effects of age, sex, pubertal status, and stage of CRF on hGH treatment.
RESULTS Literature Search We identified 337 titles and abstracts, of which 297 were excluded. The studies excluded either were clearly not clinical trials or used hGH treatment only in children with idiopathic short stature. Full assessment of the remaining 40 studies identified 12 articles that addressed the use of hGH treatment in children with CRF and provided data on clinically important outcomes.12-23 Of the 12 articles, 212,23 reported results on a subset of trial participants in another, included, study13,14 and thus were not included. Thus 10 trials involving 481 children were evaluated.
Trial Characteristics Eight of the 10 trials used a parallel design whereas the other 2 were crossover trials (Tables I and II). The trials varied in duration (6 months to 2.5 years) and included children from the United States, the United Kingdom, The Netherlands, France, and 561
VIMALACHANDRA ET AL
THE JOURNAL OF PEDIATRICS OCTOBER 2001
Table I. Characteristics of GH randomized controlled trials in children with CRF
Trial
N
Hokken-Koelega 20 (1991)15 The Netherlands Fine (1994)13 United States
125
Hokken-Koelega 11 (1996)18 The Netherlands Powell (1997)22 United States
Inclusion criteria*
Participants Children 4-16 y (mean 9.5 y) Source not stated GH: 6.0 ± 3.9 y; control: 5.7 ± 3.6 y Source: patients from 17 pediatric nephrology centers Children 8-18 y Source: not stated
Predialysis and dialysis Pubertal status: prepubertal Predialysis Pubertal status: prepubertal
Post transplantation Pubertal status: prepubertal
Intervention
Duration (mo)
GH: 28 IU/m2/wk Control: placebo
6
GH: 28 IU/m2/wk Control: placebo
24
GH: 28 IU/m2/wk Control: placebo
6
GH: 5.6 ± 2.0 y; Predialysis GH: 28 IU/m2/wk control: 5.7 ± 2.6 y Pubertal status: Control: no treatment Source: patients from prepubertal 26 participating centers 90 GH: 12.2 ± 3.1 y; Post transplantation GH: 28 IU/m2/wk Guest (1998)14 France control: 11.8 ± 3.3 y Pubertal status: Control: no treatment Source: Eighteen prepubertal and participating early pubertal French centers Children aged Post transplantation GH: 28 IU/m2/wk Maxwell (1998)21 22 United Kingdom 9.4-19.8 y Pubertal status: Control: placebo Source: 3 pediatric prepubertal nephrology centers and pubertal Hokken-Koelega 23 Children aged Predialysis and GH: 28 IU/m2/wk 16 1.6-14.1 y dialysis Control: 14 IU/m2/wk (1994) The Netherlands Source: not stated Pubertal status: prepubertal 27 Children Post transplantation GH: 28 IU/m2/wk Ito (1997)19 Japan Source: not stated Pubertal status: Control: 14 IU/m2/wk prepubertal and early pubertal Kitagawa 112 Predialysis: 8.7 ± 3.1 y; Predialysis and GH: 28 IU/m2/wk 20 dialysis: 10.1 ± 3.8 y dialysis Control: 14 IU/m2/wk (1997) Japan Source: 72 pediatric Pubertal status: kidney centers prepubertal Hokken-Koelega 18 Children aged Post transplantation GH: 56 IU/m2/wk (1994)17 11.3-19.5 y Pubertal status: early Control: 28 IU/m2/wk The Netherlands Source: not stated and late pubertal 69
12
Outcomes
1. Height velocity 2. Height velocity SDS 3. Adverse effects 4. Bone age 1. Height SDS 2. Height velocity 3. Adverse effects 4. Bone age 5. Height age 1. Height velocity 2. Height velocity SDSCA 3. Adverse effects 4. Bone age 1. Change in height SDS 2. Bone age
12
1. Height SDS 2. Height velocity 3. Adverse effects 4. Bone age
12
1. Height SDS 2. Height velocity 3. Height velocity SDS 4. Adverse effects 1. Height velocity SDS 2. Adverse effects
30
12
1. Height velocity 2. Adverse effects
12
1. Height SDS 2. Height velocity 3. Height velocity SDS 4. Adverse effects 1. Height SDSCA 2. Height velocity 3. Adverse effects
24
SDSCA, Standard deviation score chrononlogical age. *Details available on request.
Japan. At enrollment, most children were prepubertal, and some early pubertal; 1 trial included late pubertal children. Stages of CRF also varied from predialysis to posttransplanta562
tion. Four trials compared hGH (28 IU/m2/wk) with placebo, and 2 trials compared growth hormone (28 IU/m2/wk) with no treatment. The remaining 4 trials compared 2 different
doses of hGH: 3 trials compared 28 IU/m2/wk with 14 IU/m2/wk and 1 trial compared 56 IU/m2/wk with 28 IU/m 2 /wk. The type of hGH administered in the trials included
VIMALACHANDRA ET AL
THE JOURNAL OF PEDIATRICS
VOLUME 139, NUMBER 4 Table II. Quality of GH randomized controlled trials in children with CRF
Design
Method of treatment allocation
Blinding of investigators and patients
Blinded standardized outcome assessment
Hokken-Koelega15 Fine13
Crossover Parallel
Not stated Centrally randomized
Double-blind Double-blind
Yes Not stated
Hokken-Koelega18 Powell22 Guest14 Maxwell21 Hokken-Koelega16
Crossover Parallel Parallel Parallel Parallel
Not stated Not stated Centrally randomized Not stated Not stated
Double-blind Not blinded Not blinded Not stated Double-blind
Yes Not stated Not stated Not stated Yes
Trial
Ito19 Kitagawa20 Hokken-Koelega17
Parallel Parallel Parallel
Norditropin, Nutropin, Genotropin, and Genotonorm.
Not stated Not stated Not stated
Not stated Not stated Double-blind
Not stated Not stated Yes
ation in the magnitude of hGH-related growth with either pubertal status21 or stages of CRF.20
Trial Quality Quality of trial design and reporting is given in Table II. The method by which treatment assignment was blinded (allocation concealment), blinded standardized outcome assessment, and intention to treat analysis were generally not stated. Reported loss to follow-up varied from 0% to 43% between 6 months and 2.5 years.
Trial Outcomes No heterogeneity of growth outcomes within common intervention and outcome categories was demonstrated by formal testing (Tables III and IV), and so results of individual trials were pooled to obtain summary measures of treatment effects. There were no substantial differences between summary estimates calculated by use of random and fixed-effects models (fixed-effects data available on request); only results of the random effects models are reported here. Effect modification could not be formally explored because sufficient numbers of trials were not available to perform subgroup analysis or metaregression, but there was no obvious vari-
CHANGE IN HEIGHT SDS. The effect of hGH compared with placebo/no treatment on height SDS was reported in four trials (Figure). Children in the hGH group showed increased height SDS after 1 year (weighted mean difference [WMD] = 0.77 [95% CI 0.51 to 1.04]) and 2 years (1 trial,13 mean difference [MD] = 1.5 [95% CI 1.03 to 1.97]). In a child of age 10, a WMD of 0.77 approximates 4.7 cm growth in 1 year. Treatment in the second year resulted in a small and not significant increase in height SDS (1 trial,13 MD = 0.4 [95% CI –0.09 to 0.89]). Two trials comparing 2 different doses of hGH reported height SDS. Kitagawa et al 20 compared 28 IU/m 2 /wk with 14 IU/m 2 /wk and showed no difference in height SDS between the 2 groups over 6 months (MD = 0.20 [95% CI –0.33 to 0.73]) and 1 year (MD = 0.32 [95% CI –0.22 to 0.87]) (Table IV). HokkenKoelega et al17 compared 28 IU/m2/wk with 56 IU/m2/wk and also showed no difference between groups in the change in height SDS at 1 year
Follow-up (%)
Intention to treat
80 80 (1 y) 67 (2 y) 100 64 94 100 78 (1 y) 57 (2 & 2.5 y) 85 71 88 (1 y) 67 (2 y)
Not stated Not stated Yes Not stated Not stated Yes Not stated Not stated Not stated Not stated
of treatment (MD = 0.30 [95% CI –1.06 to 1.66]). CHANGE IN HEIGHT VELOCITY. Five trials compared hGH with placebo/no treatment and presented outcomes as height velocity (Table III). Two trials reported 6-month outcome data and showed an increase in height velocity of approximately 6 cm/y (WMD = 5.71 cm/y [95% CI 4.41 to 7.00]), with 2 other trials showing an increase over 1 year of 4 cm/y (WMD = 4.08 cm/y [95% CI 2.57 to 5.58]). Fine et al13 separately reported results for the second year of the study in which there was a nonsignificant decrease in height velocity in the hGH-treated group relative to the untreated group (MD = –1.90 cm/y [95% CI –3.10 to 0.70]). Three of the 4 trials comparing 2 different doses of hGH reported height velocity outcomes. Kitagawa et al20 showed an increase in height velocity to 6 months (2.78 cm/y [95% CI 1.09 to 4.46]), which waned during the second 6 months of intervention (1.39 cm/y [95% CI 0.55 to 2.23]). Ito and Kawaguchi19 reported increases of 2.7 cm/y (14 IU/m2/wk) and 2.6 cm/y (28 IU/m2/wk) in height velocity (P < .05). Hokken-Koelega et al17 compared 28 563
VIMALACHANDRA ET AL
THE JOURNAL OF PEDIATRICS OCTOBER 2001
Table III. Growth outcomes of trials (n = 6) comparing hGH with placebo/no treatment in children with CRF
Trial Hokken-Koelega15 Hokken-Koelega18 Guest14 Maxwell21 Prepubertal Pubertal Overall Q statistic
Change in height velocity Baseline to Baseline to 6 mo (cm/y) 1 y (cm/y)
Baseline to 6 mo
Baseline to 1 y
7.70* (5.28, 10.12) 7.90 (5.37, 10.43)
5.60* (3.72, 7.48) 5.80 (3.94, 7.66)
5.71 (4.41, 7.00) 0.02 (df = 1) P = .9
Change in height velocity SDS
3.20 (2.19, 4.21) 5.40† (3.21, 7.59)
6.60† (2.65, 10.55)
4.40† (1.23, 7.57) 4.08 (2.57, 5.58) 3.79 (df = 2) P = .2
5.50† (1.03, 9.97) 6.14 (3.41, 8.86) 0.15 (df = 1) P = .7
7.80 (6.09, 9.51) 0.01 (df = 1) P = .9
df, Degree of freedom. Results are given as difference in means for individual studies and weighted mean difference for the overall result with 95% CIs. The Q statistic is reported with the degrees of freedom and P value. *Data from 6 months before baseline. †Data from 1 year before baseline.
Table IV. Growth outcomes of trials comparing 28 IU/m2/wk with 14 IU/m2/wk (n = 3) of hGH treatment in children with CRF
Change in height SDS Trial
Baseline to 6 mo
Baseline to 1 y
Change in height velocity Baseline to 6 mo (cm/y)
Baseline to 1 y (cm/y)
Baseline to 6 mo
Baseline to 1 y
1.20 (0.20, 2.20) 1.90 (0.28, 3.52) 1.39 (0.55, 2.23) 0.52 (df = 1) P = .5
1.20 ( –1.73, 4.13) 1.70 (0.08, 3.32) 3.60 (1.10, 6.10) 2.05 (0.82, 3.28) 1.91 (df = 2) P = .4
0.90 ( –0.36, 2.16) 3.40 (1.02, 5.78) 1.92 ( –0.49, 4.33) 3.19 (df = 1) P = .1
Hokken-Koelega16 Kitagawa20 Predialysis Dialysis Overall Q statistic
0.20 ( –0.39, 0.79) 0.20 ( –0.87, 1.27) 0.20 ( –0.33, 0.73) 0 (df = 1) P > .9
0.30 ( –0.32, 0.92) 0.40 ( –0.67, 1.47) 0.32 ( –0.22, 0.87) 0.02 (df = 1) P = .9
2.4 (0.41, 4.39) 3.80 (0.56, 7.04) 2.78 (1.09, 4.46) 0.52 (df = 1) P = .5
Change in height velocity SDS
df, Degree of freedom. Results are given as difference in means for individual studies and weighted mean difference for the overall result with 95% CIs. The Q statistic is reported with the degrees of freedom and P value. *Data from Ito et al19 are not included in the summary estimate because no SD were provided.
IU/m2/wk with 56 IU/m2/wk and showed no significant difference in mean height velocity (1.10 cm/y [95% CI –1.30 to 3.50]) during 1 year of treatment. CHANGE IN HEIGHT VELOCITY SDS. Three trials comparing growth hormone treatment with placebo/no treatment reported data on height velocity SDS (Table III). Two trials by Hokken-Koelega et al15,18 reported an increase in height velocity SDS over 6 months’ duration (WMD = 7.80 [95% 564
CI 6.09 to 9.51]) and Maxwell and Rees21 reported an increase over 1 year of 6.14 (95% CI 3.41 to 8.86). Of the trials comparing 28 IU with 14 IU/m2/wk, 2 reported data on height velocity SDS (Table IV). At 6 months the WMD was 2.05 (95% CI 0.82 to 3.28). At 1 year Kitagawa et al20 reported an increase in height velocity SDS (WMD = 1.92 [ –0.49 to 4.33]), which was not statistically significant. CHANGE IN BONE AGE. Five trials, 13-15,18,22 all comparing growth hormone
treatment with placebo/no treatment, presented data on bone age. There was no significant difference in the change in bone age between the 2 groups over 1 year (4 trials, WMD = 0.41 [95% CI –0.40 to 1.22]) and between 1 and 2 years of treatment (1 trial, MD = 0.40 [95% CI –0.92 to 1.72]). ADVERSE EFFECTS. Reporting of adverse effects was generally scant. A number of articles reported adverse effects collectively without specifying the treatment group to which the subjects
VIMALACHANDRA ET AL
THE JOURNAL OF PEDIATRICS
VOLUME 139, NUMBER 4 Table V. Reported adverse effects of 10 trials of growth hormone treatment in children with CRF
Trials
Adverse effects
Hokken-Koelega15 Fine13
Reported “no adverse effects” Asthma/wheezing was reported in 8/55 children in the hGH group and 0/27 children from the control group. RD = 0.15 (0.01, 0.26) Reported “no acute rejection or serious deterioration during hGH therapy” None reported One patient treated with hGH had development of papilledema Eight presumed rejection episodes in 13 patients in the treatment group and in 9 patients in the control group. One patient treated with hGH had raised fasting glucose levels. One patient in the control group had worsening of a preexisting idiopathic scoliosis Reported “no adverse effects” Episodes of rejection were reported in 5/13 children in the 1.0 IU group and 2/10 children in the 0.5 IU group. RD = 0.18 ( –0.21, 0.51) More patients in the 1.0 IU than in the 0.5 IU group showed signs of deterioration in renal function. Glucose intolerance, granuloma formation at the exit site of the Tenckhoff catheter, lymph node swelling in the groin, claudication, and hypertension reported —group not specified Acute rejection and deterioration of renal function reported —group not specified
Hokken-Koelega18 Powell22 Guest14 Maxwell21
Hokken-Koelega16 Ito19 Kitagawa20
Hokken-Koelega17
Results are presented as risk difference with 95% CI, when possible.
belonged. Reported side effects included asthma/wheezing, acute rejection in transplantation, deterioration in renal function, raised fasting glucose levels, papilledema, glucose intolerance, granuloma formation, lymph node swelling, claudication, hypertension, and worsening of preexisting idiopathic scoliosis (Table V). Only 1 trial13 demonstrated a significant increase in adverse events in the hGH group relative to the control group (by use of composite end-point data). A pooled estimate of hGH-related adverse effects was not possible because of insufficient data in the primary studies.
DISCUSSION Recombinant hGH is effective in improving the growth of children with CRF during the first year of administration, with increases in all height indexes. In children treated with hGH, height SDS increases by 0.8, height velocity increases by 4 cm/y, and height velocity SDS increases by 6 above that of nontreated control subjects. It is like-
ly that such growth will not be lost in later life and represents a long-term increase in growth potential. The most important height outcome is final or adult height. Although it is possible that the final adult height of children with CRF treated with hGH is greater than non-hGH-treated children, no trials have been published with final adult height as an outcome. This is inevitable because all trials in the area have been published in the 1990s, and children recruited to these trials will only now be attaining their final adult heights. Although the final adult heights should be reported, interpretation of outcomes is likely to be difficult because of crossovers between the treated and untreated groups. Published trial data do provide some indirect support that hGH does improve final adult height in children. During 2 years of hGH treatment, there was no evidence that children grew less than their non-hGH-treated counterparts, even during the second year. It is also unlikely that hGH causes a reduction in final height, unless it acts to advance the pubertal growth spurt,
speeding up the normal growth process, and slows the growth once these children entered their usual pubertal growth spurt. This is unlikely because there was no evidence to suggest an advancement of bone age in the 5 trials in which this had been evaluated. Available randomized comparative data suggest that children with CRF should be treated with hGH 28 IU/m2/wk. Compared with a dose of 14 IU/m2/wk, the larger dose increases height by about 1.4 cm/y over 1 year, but increasing the dose to 56 IU/m2/week did not result in a statistically significant improvement in growth indices. However, these conclusions are based on only 18 patients. With the 95% CI used to quantify the uncertainty about the effect of the higher dose, relative to those patients given 28 IU/m2/wk, children given 56 IU/m2/wk may grow more than 3.5 cm/y taller in the first year or be 1.3 cm/y shorter. Across all growth outcomes, there was a consistent pattern of waning effect with longer duration of treatment. 565
VIMALACHANDRA ET AL
THE JOURNAL OF PEDIATRICS OCTOBER 2001
Figure. Change in height SDS of trials (n = 4) comparing hGH with placebo/no treatment in children with CRF. This time-dependency could not be formally tested because of insufficient data points. However, observation of the trial data is strongly suggestive of this relationship. For example, the differences in height velocity between the treated and untreated children was 5.7 cm/y (95% CI 4.4 to 7.0) for 6 months, 4.1 cm/y (95% CI 2.6 to 5.6) for the first year, and –1.9 cm/y (95% CI –3.1 to 0.7) for the second year. How long a child should be treated with hGH can be suggested for each individual patient by simply plotting height SDS and height velocity over time on a standard growth chart before and during hGH treatment to determine the effectiveness of the intervention in real-time. The benefits to growth need to be balanced with adverse events and the difficulty of adhering to a daily subcutaneous injection regimen. In a recent case series of children with CRF treated with hGH for 2 years, children who responded to hGH reported they would choose treatment again, and those who did not respond generally reported they would not choose treatment again.24 These data suggest that treatment response overrides concerns about injections. Adherence to treatment was time-dependent; thus 41% of parents reported noncompliance at 1 year, whereas 91% reported missing injections at 2 years (when re566
sponse to treatment had waned). When most parents are asked to trade off the growth potential of their children against the burden of daily injections, they opt for growth hormone treatment. Adverse event data from these randomized trials are difficult to interpret because they were underpowered to detect clinically important differences in adverse event rates and because less attention was given to the identification or reporting of adverse events than growth outcomes. In controlled trials for other indications and in observational studies of hGH in children with or without CRF, adverse events (usually minor) are reported at a frequency of about 5 per 100 patient-years.25,26 Apart from uncertainty about the duration of effective treatment, appropriate dose, and side effects of hGH, other clinically important uncertainties about the use of hGH in children with CRF remain. When should hGH administration begin? Should pubertal children be given hGH? Does the treatment effect vary according to whether the child’s status is predialysis, on hemodialysis or peritoneal dialysis, or posttransplantation? The definitive answers to these questions are not known because available trials are few and underpowered to detect differences in treatment effects across subgroups of patients on the
basis of plausible effect modifiers such as those listed. Although the trials did have some heterogeneity of participants (age, pubertal status, and stages of CRF) there is no evidence for any heterogeneity of treatment effects. Available trial data do suggest that pubertal status and treatment modality do not influence the size of response to hGH. Trial quality was frequently suboptimal, which may bias any observed treatment effects, on average, in the direction of overestimating the true treatment effect. Future trialists could improve the quality of reporting by adherence to the Consolidated Standard of Reporting Trials statement, a joint initiative by many biomedical journals to assist investigators to conduct and report randomized controlled trials in a clear and unbiased way.27 We thank Prof R. Fine, Dr L. Rees, and Ms Dechant for information on unpublished trials and Dr M. Craig for assistance with data conversion.
REFERENCES 1. Disney APS, Russ GR, Walker R, Sheil AGR, Collins J, Herbert K, et al, editors. ANZDATA Registry Report 1999, Australia and New Zealand Dialysis and Transplant Registry, Adelaide, South Australia: ANZDATA Registry; 1999.
VIMALACHANDRA ET AL
THE JOURNAL OF PEDIATRICS
VOLUME 139, NUMBER 4 2. Reynolds JM, Wood AJ, Eminson DM, Postlethwaite RJ. Short stature and chronic renal failure: what concerns children and parents. Arch Dis Child 1995;73:36-42. 3. Rizzoni G, Broyer M, Brunner FP, Brynger H, Challah S, Kramer P, et al. Combined report on regular dialysis and transplantation of children in Europe, XIII, 1983. Proc Eur Dial Transplant Assoc 1985;21:66-95. 4. Tonsholff B, Mehls O. Growth retardation in children with chronic renal insufficiency: current aspects of pathophysiology and treatment. J Nephrol 1995;8:133-42. 5. Fine RN. Recombinant human growth hormone (rhGH) in dialysis patients: update 1995. Adv Perit Dial 1995;11:261-5. 6. Clayton PE, Cowell CT. Safety issues in children and adolescents during growth hormone therapy—a review. Growth Horm IGF Res 2000;10:306-17. 7. Dickerson K, Scherer R, Lefebvre C. Identifying relevant studies for systematic reviews. In: Chalmers I, Altman DG, editors. Systematic reviews. London: BMJ Publishing Group; 1995:17-36. 8. Lefebvre C, McDonald S. Development of a sensitive search strategy for reports of randomised controlled trials in Embase. Proceedings of the Fourth International Cochrane Colloquium; Oct 20-24, 1996; Adelaide, Australia. 9. Schulz KF, Chalmers I, Hayes RJ, Altman DG. Empirical evidence of bias. Dimensions of methodological quality associated with estimates of treatment effects in controlled trials. JAMA 1995;273:408-12. 10. Neter J, Wasserman W, Kutner MH. Applied linear statistical models: regression, analysis of variance, and experimental design. Homewood, Ill: Irwin; 1990. 11. Pettiti DB. Meta-analysis decision analysis and cost-effectiveness analysis: methods for quantitative synthesis in medicine. New York: Oxford University Press; 1994.
12. Broyer M. Results and side-effects of treating children with growth hormone after kidney transplantation —a preliminary report: Pharmacia & Upjohn Study Group. Acta Paedriatr 1996; S417:76-9. 13. Fine RN, Kohaut EC, Brown D, Perlman AJ. Growth after recombinant human growth hormone treatment in children with chronic renal failure: report of a multicenter randomized double-blind placebo-controlled study— Genentech Cooperative Study Group. J Pediatr 1994;124:374-82. 14. Guest G, Berard E, Crosnier H, Chevallier T, Rappaport R, Broyer M. Effects of growth hormone in short children after renal transplantation. Pediatr Nephrol 1998;12:437-46. 15. Hokken-Koelega AC, Stijnen T, de Muinck K, Wit JM, Wolff ED, de Jong MC, et al. Placebo-controlled, double-blind, cross-over trial of growth hormone treatment in prepubertal children with chronic renal failure. Lancet 1991;338:585-90. 16. Hokken-Koelega AC, Stijnen T, De Jong MC, Donckerwolcke RA, de Muinck Keizer-Schrama SM, Blum WF, et al. Double blind trial comparing the effects of two doses of growth hormone in prepubertal patients with chronic renal insufficiency. J Clin Endocrinol Metab 1994;79:1185-90. 17. Hokken-Koelega AC, Stijnen T, de Ridder MA, de Muinck Keizer-Schrama SM, Wolff ED, de Jong MC, et al. Growth hormone treatment in growthretarded adolescents after renal transplant. Lancet 1994;343:1313-7. 18. Hokken-Koelega AC, Stijnen T, de Jong RC, Donckerwolcke RA, Groothoff JW, Wolff ED et al. A placebo-controlled, double-blind trial of growth hormone treatment in prepubertal children after renal transplant. Kidney Int 1996;53S:S128-34. 19. Ito K, Kawaguchi H. The use of recombinant human GH (rhGH) in growth
20.
21.
22.
23.
24.
25.
26.
27.
failure in children with various renal diseases. Clin Pediatr Endocrinol 1997; S9:49-53. Kitagawa T, Ito K, Ito H, Sakai T, Wada H, Kajiwara N. GH treatment of children with chronic renal insufficiency: a Japanese clinical trial. Clin Pediatr Endocrinol 1997;S10:73-80. Maxwell H, Rees L. Randomised controlled trial of recombinant human growth hormone in prepubertal and pubertal renal transplant recipients. Arch Dis Child 1998;79:181-7. Powell DR, Liu F, Baker BK, Hintz RL, Lee PD, Durham SK, et al. Modulation of growth factors by growth hormone in children with chronic renal failure: the Southwest Pediatric Nephrology Study Group. Kidney Int 1997;51:1970-9. Fine RN, Attie KM, Kuntze J, Brown DF, Kohaut EC. Recombinant human growth hormone in infants and young children with chronic renal insufficiency: Genentech Collaborative Study Group. Pediatr Nephrol 1995;9:451-7. Postlethwaite RJ, Eminson DM, Reynolds JM, Wood AJ, Hollis S. Growth in renal failure: a longitudinal study of emotional and behavioural changes during trials of growth hormone treatment. Arch Dis Child 1998;78:222-9. Cowell CT, Dietsch S. Adverse events during growth hormone therapy. J Pediatr Endocrinol Metab 1995; 8:243-52. Wilton P. Adverse events during GH treatment: 10 years’ experience in KIGS, a pharmacoepidemiological survey. In: Ranke MB, Wilton P, editors. Progress in growth hormone therapy – 10 years of KIGS. Mannheim: Ja Barh Verlag; 1999. p. 349-64. Begg C, Cho M, Eastwood S, Horton R, Moher D, Olkin I, et al. Improving the quality of reporting of randomized controlled trials: the CONSORT statement. JAMA 1996;276:637-9.
567
As defined by the need for dialysis or kidney transplantation, 32 per 1 million children younger than 15 years have chronic renal failure (CRF)1; many
more children have less advanced CRF (glomerular filtration rate [GFR] < 30 mL/min/1.73 m2). Growth retardation, one of the complications of CRF, is of
From the Centre for Kidney Research, The Children’s Hospital at Westmead, Robert Vines Growth Research Centre, The Children’s Hospital at Westmead, and the Department of Public Health and Community Medicine, University of Sydney, Australia.
Supported by the Australian Kidney Foundation, the Small Grants Scheme of the Royal Alexandra Hospital for Children, and the Cochrane Child Health Field Bursary. Submitted for publication Aug 8, 2000; revisions received Dec 21, 2000, and Feb 20, 2001; accepted May 25, 2001. Reprint requests: Dushyanthi Vimalachandra, Centre for Kidney Research, The Children’s Hospital at Westmead, Locked Bag 4001, Westmead, NSW 2145, Australia. Copyright © 2001 by Mosby, Inc. 0022-3476/2001/$35.00 + 0 9/21/117582 doi:10.1067/mpd.2001.117582
560
concern to the families of more than 90% of children with CRF.2 Approximately 60% of boys and 41% of girls who started renal replacement therapy before age 15 years attain a final adult height more than 2 SD (12 cm) below the mean for healthy adults.3 Impairment of growth can begin when the GFR falls to 50% of normal and becomes an increasing problem once the GFR falls below 25%.4 Over the past 10 years, recombinant CRF GFR hGH MD SDS WMD
Chronic renal failure Glomerular filtration rate Human growth hormone Mean difference Standard deviation score Weighted mean difference
human growth hormone (hGH) treatment has been used to help short children with CRF attain a height more in keeping with their age group.5 However, there are concerns that hGH may have an adverse effect on the preservation of native renal function, predispose to acute rejection in kidney transplant recipients, and cause benign intracranial hypertension and slipped capital femoral epiphysis.6 Although many trials of hGH treatment in children with CRF have been undertaken, uncertainty exists on the magnitude of benefits and side effects of the treatment.5 A systematic review of hGH treatment was undertaken to evaluate growth outcomes to establish the effect of treatment over time. We sought to establish whether the growth outcomes remained linear over time or whether there was a waning effect of
VIMALACHANDRA ET AL
THE JOURNAL OF PEDIATRICS
VOLUME 139, NUMBER 4 the treatment. Second, we examined the effect of varying doses of the treatment. Third, we explored the effect of the following factors on treatment: age, sex, pubertal status, and the stage of CRF (before dialysis, on dialysis, after transplantation). Finally, the study evaluated potential side effects of hGH treatment.
METHODS Inclusion Criteria Randomized controlled trials that compared hGH treatment with placebo or no treatment or that compared 2 doses of hGH treatment were identified. The study participants were under age 18 years, diagnosed with renal failure and either had not yet started dialysis, were on dialysis, or had undergone transplantation. Growth outcomes were assessed by standard deviation score (SDS). An individual’s height SDS was calculated as follows: (Actual height – Mean height for age)/SD at that age. Normal children have a height that ranges between –2 and +2 SDS. If a change in height SDS is negative, it implies growth failure and if positive, catch-up growth. The primary outcome measure was the difference in mean change in height SDS between the treatment and control groups during the trial period. This was computed as [(postmean height SDS(treated) – pre-mean height SDS (treated) ) – (post-mean height SDS(control) – pre-mean height SDS (control) )]. Secondary outcomes included change in height SDS from treatment onset to completion, change in height SDS during puberty, change in height velocity, change in height velocity SDS, final height, change in bone age, quality of life, and adverse effects of hGH treatment (hypertension, graft function, rejection episodes, and any other possible side effects reported).
Search Strategy A systematic and comprehensive literature search was carried out to identify eligible randomized controlled trials.
The Cochrane Controlled Trials Register (Issue 2, 2000) was searched with the following search terms used as medical subject heading and text words: growth hormone/somatotropin, chronic renal failure, end stage renal disease, and kidney transplantation. Two reviewers (D.V. and J.K.) independently searched Medline (1966 to April 2000) and Embase (1988 to April 2000) without language restriction. An optimally sensitive search strategy developed for identification of randomized controlled trials7,8 was combined with medical subject heading and text words for CRF and growth hormone. Additional studies were located through article reference lists and through contact with local and international experts in the field. Two reviewers (D.V. and J.K.) independently screened titles and abstracts of studies identified through the searches and selected trials that met the inclusion criteria. This process favored overselection to avoid losing relevant studies. In the event of uncertainty or no abstract available, the full article was obtained. Any disagreement on study selection was resolved through discussion and consultation with a third reviewer (J.C.).
Data Extraction and Quality Assessment Two reviewers (D.V. and J.K.) independently extracted data from eligible full articles. Information was collected on participant characteristics (number, age, sex, pubertal stage, stage of CRF, comorbidities), intervention (type of growth hormone, dose, duration, cointerventions), and primary and secondary outcome measures. Methodologic quality (allocation concealment, blinding of outcome assessment, completeness of follow-up, and analysis by intention to treat9) of the eligible studies were also assessed. In the case of cross-over trials, data from the first half of the study were extracted and used in the analysis. Authors were contacted to obtain any missing data, as well as raw data when necessary. Any discrepancies in data extraction were discussed with a third re-
viewer (J.C.). In the case where the results of a study were published more than once or the results were discussed in a number of publications, the most complete data were obtained by use of all sources and included in the analysis only once.
Statistical Analysis Difference in means was calculated for continuous outcomes of individual trials to estimate the size of the intervention effects. In the case where a trial reported only 6-month velocity data, the mean and SD were doubled to arrive at annualized data.10 Overall treatment effects (weighted mean difference) were computed by use of a random effects model that takes into account betweenstudy variability, as well as within-study variability. Heterogeneity was analyzed with Cochran’s Q statistic.11 Subgroup analysis was attempted to explore the effects of age, sex, pubertal status, and stage of CRF on hGH treatment.
RESULTS Literature Search We identified 337 titles and abstracts, of which 297 were excluded. The studies excluded either were clearly not clinical trials or used hGH treatment only in children with idiopathic short stature. Full assessment of the remaining 40 studies identified 12 articles that addressed the use of hGH treatment in children with CRF and provided data on clinically important outcomes.12-23 Of the 12 articles, 212,23 reported results on a subset of trial participants in another, included, study13,14 and thus were not included. Thus 10 trials involving 481 children were evaluated.
Trial Characteristics Eight of the 10 trials used a parallel design whereas the other 2 were crossover trials (Tables I and II). The trials varied in duration (6 months to 2.5 years) and included children from the United States, the United Kingdom, The Netherlands, France, and 561
VIMALACHANDRA ET AL
THE JOURNAL OF PEDIATRICS OCTOBER 2001
Table I. Characteristics of GH randomized controlled trials in children with CRF
Trial
N
Hokken-Koelega 20 (1991)15 The Netherlands Fine (1994)13 United States
125
Hokken-Koelega 11 (1996)18 The Netherlands Powell (1997)22 United States
Inclusion criteria*
Participants Children 4-16 y (mean 9.5 y) Source not stated GH: 6.0 ± 3.9 y; control: 5.7 ± 3.6 y Source: patients from 17 pediatric nephrology centers Children 8-18 y Source: not stated
Predialysis and dialysis Pubertal status: prepubertal Predialysis Pubertal status: prepubertal
Post transplantation Pubertal status: prepubertal
Intervention
Duration (mo)
GH: 28 IU/m2/wk Control: placebo
6
GH: 28 IU/m2/wk Control: placebo
24
GH: 28 IU/m2/wk Control: placebo
6
GH: 5.6 ± 2.0 y; Predialysis GH: 28 IU/m2/wk control: 5.7 ± 2.6 y Pubertal status: Control: no treatment Source: patients from prepubertal 26 participating centers 90 GH: 12.2 ± 3.1 y; Post transplantation GH: 28 IU/m2/wk Guest (1998)14 France control: 11.8 ± 3.3 y Pubertal status: Control: no treatment Source: Eighteen prepubertal and participating early pubertal French centers Children aged Post transplantation GH: 28 IU/m2/wk Maxwell (1998)21 22 United Kingdom 9.4-19.8 y Pubertal status: Control: placebo Source: 3 pediatric prepubertal nephrology centers and pubertal Hokken-Koelega 23 Children aged Predialysis and GH: 28 IU/m2/wk 16 1.6-14.1 y dialysis Control: 14 IU/m2/wk (1994) The Netherlands Source: not stated Pubertal status: prepubertal 27 Children Post transplantation GH: 28 IU/m2/wk Ito (1997)19 Japan Source: not stated Pubertal status: Control: 14 IU/m2/wk prepubertal and early pubertal Kitagawa 112 Predialysis: 8.7 ± 3.1 y; Predialysis and GH: 28 IU/m2/wk 20 dialysis: 10.1 ± 3.8 y dialysis Control: 14 IU/m2/wk (1997) Japan Source: 72 pediatric Pubertal status: kidney centers prepubertal Hokken-Koelega 18 Children aged Post transplantation GH: 56 IU/m2/wk (1994)17 11.3-19.5 y Pubertal status: early Control: 28 IU/m2/wk The Netherlands Source: not stated and late pubertal 69
12
Outcomes
1. Height velocity 2. Height velocity SDS 3. Adverse effects 4. Bone age 1. Height SDS 2. Height velocity 3. Adverse effects 4. Bone age 5. Height age 1. Height velocity 2. Height velocity SDSCA 3. Adverse effects 4. Bone age 1. Change in height SDS 2. Bone age
12
1. Height SDS 2. Height velocity 3. Adverse effects 4. Bone age
12
1. Height SDS 2. Height velocity 3. Height velocity SDS 4. Adverse effects 1. Height velocity SDS 2. Adverse effects
30
12
1. Height velocity 2. Adverse effects
12
1. Height SDS 2. Height velocity 3. Height velocity SDS 4. Adverse effects 1. Height SDSCA 2. Height velocity 3. Adverse effects
24
SDSCA, Standard deviation score chrononlogical age. *Details available on request.
Japan. At enrollment, most children were prepubertal, and some early pubertal; 1 trial included late pubertal children. Stages of CRF also varied from predialysis to posttransplanta562
tion. Four trials compared hGH (28 IU/m2/wk) with placebo, and 2 trials compared growth hormone (28 IU/m2/wk) with no treatment. The remaining 4 trials compared 2 different
doses of hGH: 3 trials compared 28 IU/m2/wk with 14 IU/m2/wk and 1 trial compared 56 IU/m2/wk with 28 IU/m 2 /wk. The type of hGH administered in the trials included
VIMALACHANDRA ET AL
THE JOURNAL OF PEDIATRICS
VOLUME 139, NUMBER 4 Table II. Quality of GH randomized controlled trials in children with CRF
Design
Method of treatment allocation
Blinding of investigators and patients
Blinded standardized outcome assessment
Hokken-Koelega15 Fine13
Crossover Parallel
Not stated Centrally randomized
Double-blind Double-blind
Yes Not stated
Hokken-Koelega18 Powell22 Guest14 Maxwell21 Hokken-Koelega16
Crossover Parallel Parallel Parallel Parallel
Not stated Not stated Centrally randomized Not stated Not stated
Double-blind Not blinded Not blinded Not stated Double-blind
Yes Not stated Not stated Not stated Yes
Trial
Ito19 Kitagawa20 Hokken-Koelega17
Parallel Parallel Parallel
Norditropin, Nutropin, Genotropin, and Genotonorm.
Not stated Not stated Not stated
Not stated Not stated Double-blind
Not stated Not stated Yes
ation in the magnitude of hGH-related growth with either pubertal status21 or stages of CRF.20
Trial Quality Quality of trial design and reporting is given in Table II. The method by which treatment assignment was blinded (allocation concealment), blinded standardized outcome assessment, and intention to treat analysis were generally not stated. Reported loss to follow-up varied from 0% to 43% between 6 months and 2.5 years.
Trial Outcomes No heterogeneity of growth outcomes within common intervention and outcome categories was demonstrated by formal testing (Tables III and IV), and so results of individual trials were pooled to obtain summary measures of treatment effects. There were no substantial differences between summary estimates calculated by use of random and fixed-effects models (fixed-effects data available on request); only results of the random effects models are reported here. Effect modification could not be formally explored because sufficient numbers of trials were not available to perform subgroup analysis or metaregression, but there was no obvious vari-
CHANGE IN HEIGHT SDS. The effect of hGH compared with placebo/no treatment on height SDS was reported in four trials (Figure). Children in the hGH group showed increased height SDS after 1 year (weighted mean difference [WMD] = 0.77 [95% CI 0.51 to 1.04]) and 2 years (1 trial,13 mean difference [MD] = 1.5 [95% CI 1.03 to 1.97]). In a child of age 10, a WMD of 0.77 approximates 4.7 cm growth in 1 year. Treatment in the second year resulted in a small and not significant increase in height SDS (1 trial,13 MD = 0.4 [95% CI –0.09 to 0.89]). Two trials comparing 2 different doses of hGH reported height SDS. Kitagawa et al 20 compared 28 IU/m 2 /wk with 14 IU/m 2 /wk and showed no difference in height SDS between the 2 groups over 6 months (MD = 0.20 [95% CI –0.33 to 0.73]) and 1 year (MD = 0.32 [95% CI –0.22 to 0.87]) (Table IV). HokkenKoelega et al17 compared 28 IU/m2/wk with 56 IU/m2/wk and also showed no difference between groups in the change in height SDS at 1 year
Follow-up (%)
Intention to treat
80 80 (1 y) 67 (2 y) 100 64 94 100 78 (1 y) 57 (2 & 2.5 y) 85 71 88 (1 y) 67 (2 y)
Not stated Not stated Yes Not stated Not stated Yes Not stated Not stated Not stated Not stated
of treatment (MD = 0.30 [95% CI –1.06 to 1.66]). CHANGE IN HEIGHT VELOCITY. Five trials compared hGH with placebo/no treatment and presented outcomes as height velocity (Table III). Two trials reported 6-month outcome data and showed an increase in height velocity of approximately 6 cm/y (WMD = 5.71 cm/y [95% CI 4.41 to 7.00]), with 2 other trials showing an increase over 1 year of 4 cm/y (WMD = 4.08 cm/y [95% CI 2.57 to 5.58]). Fine et al13 separately reported results for the second year of the study in which there was a nonsignificant decrease in height velocity in the hGH-treated group relative to the untreated group (MD = –1.90 cm/y [95% CI –3.10 to 0.70]). Three of the 4 trials comparing 2 different doses of hGH reported height velocity outcomes. Kitagawa et al20 showed an increase in height velocity to 6 months (2.78 cm/y [95% CI 1.09 to 4.46]), which waned during the second 6 months of intervention (1.39 cm/y [95% CI 0.55 to 2.23]). Ito and Kawaguchi19 reported increases of 2.7 cm/y (14 IU/m2/wk) and 2.6 cm/y (28 IU/m2/wk) in height velocity (P < .05). Hokken-Koelega et al17 compared 28 563
VIMALACHANDRA ET AL
THE JOURNAL OF PEDIATRICS OCTOBER 2001
Table III. Growth outcomes of trials (n = 6) comparing hGH with placebo/no treatment in children with CRF
Trial Hokken-Koelega15 Hokken-Koelega18 Guest14 Maxwell21 Prepubertal Pubertal Overall Q statistic
Change in height velocity Baseline to Baseline to 6 mo (cm/y) 1 y (cm/y)
Baseline to 6 mo
Baseline to 1 y
7.70* (5.28, 10.12) 7.90 (5.37, 10.43)
5.60* (3.72, 7.48) 5.80 (3.94, 7.66)
5.71 (4.41, 7.00) 0.02 (df = 1) P = .9
Change in height velocity SDS
3.20 (2.19, 4.21) 5.40† (3.21, 7.59)
6.60† (2.65, 10.55)
4.40† (1.23, 7.57) 4.08 (2.57, 5.58) 3.79 (df = 2) P = .2
5.50† (1.03, 9.97) 6.14 (3.41, 8.86) 0.15 (df = 1) P = .7
7.80 (6.09, 9.51) 0.01 (df = 1) P = .9
df, Degree of freedom. Results are given as difference in means for individual studies and weighted mean difference for the overall result with 95% CIs. The Q statistic is reported with the degrees of freedom and P value. *Data from 6 months before baseline. †Data from 1 year before baseline.
Table IV. Growth outcomes of trials comparing 28 IU/m2/wk with 14 IU/m2/wk (n = 3) of hGH treatment in children with CRF
Change in height SDS Trial
Baseline to 6 mo
Baseline to 1 y
Change in height velocity Baseline to 6 mo (cm/y)
Baseline to 1 y (cm/y)
Baseline to 6 mo
Baseline to 1 y
1.20 (0.20, 2.20) 1.90 (0.28, 3.52) 1.39 (0.55, 2.23) 0.52 (df = 1) P = .5
1.20 ( –1.73, 4.13) 1.70 (0.08, 3.32) 3.60 (1.10, 6.10) 2.05 (0.82, 3.28) 1.91 (df = 2) P = .4
0.90 ( –0.36, 2.16) 3.40 (1.02, 5.78) 1.92 ( –0.49, 4.33) 3.19 (df = 1) P = .1
Hokken-Koelega16 Kitagawa20 Predialysis Dialysis Overall Q statistic
0.20 ( –0.39, 0.79) 0.20 ( –0.87, 1.27) 0.20 ( –0.33, 0.73) 0 (df = 1) P > .9
0.30 ( –0.32, 0.92) 0.40 ( –0.67, 1.47) 0.32 ( –0.22, 0.87) 0.02 (df = 1) P = .9
2.4 (0.41, 4.39) 3.80 (0.56, 7.04) 2.78 (1.09, 4.46) 0.52 (df = 1) P = .5
Change in height velocity SDS
df, Degree of freedom. Results are given as difference in means for individual studies and weighted mean difference for the overall result with 95% CIs. The Q statistic is reported with the degrees of freedom and P value. *Data from Ito et al19 are not included in the summary estimate because no SD were provided.
IU/m2/wk with 56 IU/m2/wk and showed no significant difference in mean height velocity (1.10 cm/y [95% CI –1.30 to 3.50]) during 1 year of treatment. CHANGE IN HEIGHT VELOCITY SDS. Three trials comparing growth hormone treatment with placebo/no treatment reported data on height velocity SDS (Table III). Two trials by Hokken-Koelega et al15,18 reported an increase in height velocity SDS over 6 months’ duration (WMD = 7.80 [95% 564
CI 6.09 to 9.51]) and Maxwell and Rees21 reported an increase over 1 year of 6.14 (95% CI 3.41 to 8.86). Of the trials comparing 28 IU with 14 IU/m2/wk, 2 reported data on height velocity SDS (Table IV). At 6 months the WMD was 2.05 (95% CI 0.82 to 3.28). At 1 year Kitagawa et al20 reported an increase in height velocity SDS (WMD = 1.92 [ –0.49 to 4.33]), which was not statistically significant. CHANGE IN BONE AGE. Five trials, 13-15,18,22 all comparing growth hormone
treatment with placebo/no treatment, presented data on bone age. There was no significant difference in the change in bone age between the 2 groups over 1 year (4 trials, WMD = 0.41 [95% CI –0.40 to 1.22]) and between 1 and 2 years of treatment (1 trial, MD = 0.40 [95% CI –0.92 to 1.72]). ADVERSE EFFECTS. Reporting of adverse effects was generally scant. A number of articles reported adverse effects collectively without specifying the treatment group to which the subjects
VIMALACHANDRA ET AL
THE JOURNAL OF PEDIATRICS
VOLUME 139, NUMBER 4 Table V. Reported adverse effects of 10 trials of growth hormone treatment in children with CRF
Trials
Adverse effects
Hokken-Koelega15 Fine13
Reported “no adverse effects” Asthma/wheezing was reported in 8/55 children in the hGH group and 0/27 children from the control group. RD = 0.15 (0.01, 0.26) Reported “no acute rejection or serious deterioration during hGH therapy” None reported One patient treated with hGH had development of papilledema Eight presumed rejection episodes in 13 patients in the treatment group and in 9 patients in the control group. One patient treated with hGH had raised fasting glucose levels. One patient in the control group had worsening of a preexisting idiopathic scoliosis Reported “no adverse effects” Episodes of rejection were reported in 5/13 children in the 1.0 IU group and 2/10 children in the 0.5 IU group. RD = 0.18 ( –0.21, 0.51) More patients in the 1.0 IU than in the 0.5 IU group showed signs of deterioration in renal function. Glucose intolerance, granuloma formation at the exit site of the Tenckhoff catheter, lymph node swelling in the groin, claudication, and hypertension reported —group not specified Acute rejection and deterioration of renal function reported —group not specified
Hokken-Koelega18 Powell22 Guest14 Maxwell21
Hokken-Koelega16 Ito19 Kitagawa20
Hokken-Koelega17
Results are presented as risk difference with 95% CI, when possible.
belonged. Reported side effects included asthma/wheezing, acute rejection in transplantation, deterioration in renal function, raised fasting glucose levels, papilledema, glucose intolerance, granuloma formation, lymph node swelling, claudication, hypertension, and worsening of preexisting idiopathic scoliosis (Table V). Only 1 trial13 demonstrated a significant increase in adverse events in the hGH group relative to the control group (by use of composite end-point data). A pooled estimate of hGH-related adverse effects was not possible because of insufficient data in the primary studies.
DISCUSSION Recombinant hGH is effective in improving the growth of children with CRF during the first year of administration, with increases in all height indexes. In children treated with hGH, height SDS increases by 0.8, height velocity increases by 4 cm/y, and height velocity SDS increases by 6 above that of nontreated control subjects. It is like-
ly that such growth will not be lost in later life and represents a long-term increase in growth potential. The most important height outcome is final or adult height. Although it is possible that the final adult height of children with CRF treated with hGH is greater than non-hGH-treated children, no trials have been published with final adult height as an outcome. This is inevitable because all trials in the area have been published in the 1990s, and children recruited to these trials will only now be attaining their final adult heights. Although the final adult heights should be reported, interpretation of outcomes is likely to be difficult because of crossovers between the treated and untreated groups. Published trial data do provide some indirect support that hGH does improve final adult height in children. During 2 years of hGH treatment, there was no evidence that children grew less than their non-hGH-treated counterparts, even during the second year. It is also unlikely that hGH causes a reduction in final height, unless it acts to advance the pubertal growth spurt,
speeding up the normal growth process, and slows the growth once these children entered their usual pubertal growth spurt. This is unlikely because there was no evidence to suggest an advancement of bone age in the 5 trials in which this had been evaluated. Available randomized comparative data suggest that children with CRF should be treated with hGH 28 IU/m2/wk. Compared with a dose of 14 IU/m2/wk, the larger dose increases height by about 1.4 cm/y over 1 year, but increasing the dose to 56 IU/m2/week did not result in a statistically significant improvement in growth indices. However, these conclusions are based on only 18 patients. With the 95% CI used to quantify the uncertainty about the effect of the higher dose, relative to those patients given 28 IU/m2/wk, children given 56 IU/m2/wk may grow more than 3.5 cm/y taller in the first year or be 1.3 cm/y shorter. Across all growth outcomes, there was a consistent pattern of waning effect with longer duration of treatment. 565
VIMALACHANDRA ET AL
THE JOURNAL OF PEDIATRICS OCTOBER 2001
Figure. Change in height SDS of trials (n = 4) comparing hGH with placebo/no treatment in children with CRF. This time-dependency could not be formally tested because of insufficient data points. However, observation of the trial data is strongly suggestive of this relationship. For example, the differences in height velocity between the treated and untreated children was 5.7 cm/y (95% CI 4.4 to 7.0) for 6 months, 4.1 cm/y (95% CI 2.6 to 5.6) for the first year, and –1.9 cm/y (95% CI –3.1 to 0.7) for the second year. How long a child should be treated with hGH can be suggested for each individual patient by simply plotting height SDS and height velocity over time on a standard growth chart before and during hGH treatment to determine the effectiveness of the intervention in real-time. The benefits to growth need to be balanced with adverse events and the difficulty of adhering to a daily subcutaneous injection regimen. In a recent case series of children with CRF treated with hGH for 2 years, children who responded to hGH reported they would choose treatment again, and those who did not respond generally reported they would not choose treatment again.24 These data suggest that treatment response overrides concerns about injections. Adherence to treatment was time-dependent; thus 41% of parents reported noncompliance at 1 year, whereas 91% reported missing injections at 2 years (when re566
sponse to treatment had waned). When most parents are asked to trade off the growth potential of their children against the burden of daily injections, they opt for growth hormone treatment. Adverse event data from these randomized trials are difficult to interpret because they were underpowered to detect clinically important differences in adverse event rates and because less attention was given to the identification or reporting of adverse events than growth outcomes. In controlled trials for other indications and in observational studies of hGH in children with or without CRF, adverse events (usually minor) are reported at a frequency of about 5 per 100 patient-years.25,26 Apart from uncertainty about the duration of effective treatment, appropriate dose, and side effects of hGH, other clinically important uncertainties about the use of hGH in children with CRF remain. When should hGH administration begin? Should pubertal children be given hGH? Does the treatment effect vary according to whether the child’s status is predialysis, on hemodialysis or peritoneal dialysis, or posttransplantation? The definitive answers to these questions are not known because available trials are few and underpowered to detect differences in treatment effects across subgroups of patients on the
basis of plausible effect modifiers such as those listed. Although the trials did have some heterogeneity of participants (age, pubertal status, and stages of CRF) there is no evidence for any heterogeneity of treatment effects. Available trial data do suggest that pubertal status and treatment modality do not influence the size of response to hGH. Trial quality was frequently suboptimal, which may bias any observed treatment effects, on average, in the direction of overestimating the true treatment effect. Future trialists could improve the quality of reporting by adherence to the Consolidated Standard of Reporting Trials statement, a joint initiative by many biomedical journals to assist investigators to conduct and report randomized controlled trials in a clear and unbiased way.27 We thank Prof R. Fine, Dr L. Rees, and Ms Dechant for information on unpublished trials and Dr M. Craig for assistance with data conversion.
REFERENCES 1. Disney APS, Russ GR, Walker R, Sheil AGR, Collins J, Herbert K, et al, editors. ANZDATA Registry Report 1999, Australia and New Zealand Dialysis and Transplant Registry, Adelaide, South Australia: ANZDATA Registry; 1999.
VIMALACHANDRA ET AL
THE JOURNAL OF PEDIATRICS
VOLUME 139, NUMBER 4 2. Reynolds JM, Wood AJ, Eminson DM, Postlethwaite RJ. Short stature and chronic renal failure: what concerns children and parents. Arch Dis Child 1995;73:36-42. 3. Rizzoni G, Broyer M, Brunner FP, Brynger H, Challah S, Kramer P, et al. Combined report on regular dialysis and transplantation of children in Europe, XIII, 1983. Proc Eur Dial Transplant Assoc 1985;21:66-95. 4. Tonsholff B, Mehls O. Growth retardation in children with chronic renal insufficiency: current aspects of pathophysiology and treatment. J Nephrol 1995;8:133-42. 5. Fine RN. Recombinant human growth hormone (rhGH) in dialysis patients: update 1995. Adv Perit Dial 1995;11:261-5. 6. Clayton PE, Cowell CT. Safety issues in children and adolescents during growth hormone therapy—a review. Growth Horm IGF Res 2000;10:306-17. 7. Dickerson K, Scherer R, Lefebvre C. Identifying relevant studies for systematic reviews. In: Chalmers I, Altman DG, editors. Systematic reviews. London: BMJ Publishing Group; 1995:17-36. 8. Lefebvre C, McDonald S. Development of a sensitive search strategy for reports of randomised controlled trials in Embase. Proceedings of the Fourth International Cochrane Colloquium; Oct 20-24, 1996; Adelaide, Australia. 9. Schulz KF, Chalmers I, Hayes RJ, Altman DG. Empirical evidence of bias. Dimensions of methodological quality associated with estimates of treatment effects in controlled trials. JAMA 1995;273:408-12. 10. Neter J, Wasserman W, Kutner MH. Applied linear statistical models: regression, analysis of variance, and experimental design. Homewood, Ill: Irwin; 1990. 11. Pettiti DB. Meta-analysis decision analysis and cost-effectiveness analysis: methods for quantitative synthesis in medicine. New York: Oxford University Press; 1994.
12. Broyer M. Results and side-effects of treating children with growth hormone after kidney transplantation —a preliminary report: Pharmacia & Upjohn Study Group. Acta Paedriatr 1996; S417:76-9. 13. Fine RN, Kohaut EC, Brown D, Perlman AJ. Growth after recombinant human growth hormone treatment in children with chronic renal failure: report of a multicenter randomized double-blind placebo-controlled study— Genentech Cooperative Study Group. J Pediatr 1994;124:374-82. 14. Guest G, Berard E, Crosnier H, Chevallier T, Rappaport R, Broyer M. Effects of growth hormone in short children after renal transplantation. Pediatr Nephrol 1998;12:437-46. 15. Hokken-Koelega AC, Stijnen T, de Muinck K, Wit JM, Wolff ED, de Jong MC, et al. Placebo-controlled, double-blind, cross-over trial of growth hormone treatment in prepubertal children with chronic renal failure. Lancet 1991;338:585-90. 16. Hokken-Koelega AC, Stijnen T, De Jong MC, Donckerwolcke RA, de Muinck Keizer-Schrama SM, Blum WF, et al. Double blind trial comparing the effects of two doses of growth hormone in prepubertal patients with chronic renal insufficiency. J Clin Endocrinol Metab 1994;79:1185-90. 17. Hokken-Koelega AC, Stijnen T, de Ridder MA, de Muinck Keizer-Schrama SM, Wolff ED, de Jong MC, et al. Growth hormone treatment in growthretarded adolescents after renal transplant. Lancet 1994;343:1313-7. 18. Hokken-Koelega AC, Stijnen T, de Jong RC, Donckerwolcke RA, Groothoff JW, Wolff ED et al. A placebo-controlled, double-blind trial of growth hormone treatment in prepubertal children after renal transplant. Kidney Int 1996;53S:S128-34. 19. Ito K, Kawaguchi H. The use of recombinant human GH (rhGH) in growth
20.
21.
22.
23.
24.
25.
26.
27.
failure in children with various renal diseases. Clin Pediatr Endocrinol 1997; S9:49-53. Kitagawa T, Ito K, Ito H, Sakai T, Wada H, Kajiwara N. GH treatment of children with chronic renal insufficiency: a Japanese clinical trial. Clin Pediatr Endocrinol 1997;S10:73-80. Maxwell H, Rees L. Randomised controlled trial of recombinant human growth hormone in prepubertal and pubertal renal transplant recipients. Arch Dis Child 1998;79:181-7. Powell DR, Liu F, Baker BK, Hintz RL, Lee PD, Durham SK, et al. Modulation of growth factors by growth hormone in children with chronic renal failure: the Southwest Pediatric Nephrology Study Group. Kidney Int 1997;51:1970-9. Fine RN, Attie KM, Kuntze J, Brown DF, Kohaut EC. Recombinant human growth hormone in infants and young children with chronic renal insufficiency: Genentech Collaborative Study Group. Pediatr Nephrol 1995;9:451-7. Postlethwaite RJ, Eminson DM, Reynolds JM, Wood AJ, Hollis S. Growth in renal failure: a longitudinal study of emotional and behavioural changes during trials of growth hormone treatment. Arch Dis Child 1998;78:222-9. Cowell CT, Dietsch S. Adverse events during growth hormone therapy. J Pediatr Endocrinol Metab 1995; 8:243-52. Wilton P. Adverse events during GH treatment: 10 years’ experience in KIGS, a pharmacoepidemiological survey. In: Ranke MB, Wilton P, editors. Progress in growth hormone therapy – 10 years of KIGS. Mannheim: Ja Barh Verlag; 1999. p. 349-64. Begg C, Cho M, Eastwood S, Horton R, Moher D, Olkin I, et al. Improving the quality of reporting of randomized controlled trials: the CONSORT statement. JAMA 1996;276:637-9.
567