PubMed Health. A service of the National Library of Medicine, National Institutes of Health.

Phung OJ, Coleman CI, Baker EL, et al. Effectiveness of Recombinant Human Growth Hormone (rhGH) in the Treatment of Patients With Cystic Fibrosis [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2010 Oct. (Comparative Effectiveness Reviews, No. 23.)

3Results

Results of Literature Search

A summary of search results is presented in Figure 2Figure 4.

Figure 2. PRISMA flow diagram of search for KQs 1,2,4,6,7.

Figure 2

PRISMA flow diagram of search for KQs 1,2,4,6,7. Legend: CF=cystic fibrosis; KQ=key question; PRISMA=preferred reporting items for systematic reviews and meta-analyses; rhGH=recombinant human growth hormone Narrative: Figure 2 shows the flow of study (more...)

Figure 3. PRISMA flow diagram of search for KQ 3.

Figure 3

PRISMA flow diagram of search for KQ 3. Legend: CF=cystic fibrosis; KQ=key question; PRISMA=preferred reporting items for systematic reviews and meta-analyses Narrative: Figure 3 shows the flow of study identification and selection. The original database (more...)

Figure 4. PRISMA flow diagram of search for KQ 5.

Figure 4

PRISMA flow diagram of search for KQ 5. Legend: CF=cystic fibrosis; GHD=growth hormone deficiency; ISS=idiopathic short stature; KQ=key question; PRISMA=preferred reporting items for systematic reviews and meta-analyses; rhGH=recombinant human growth (more...)

Upon conducting the literature search to identify articles that evaluated the use of rhGH in CF populations, we retrieved 44 unique citations and another citation was indentified from other sources. Eighteen articles were excluded during the title and abstract review and two articles were excluded during the full text review. A total of 26 articles were found to match our inclusion criteria.4,16,24–47

From the literature search for studies which evaluated the linkages between intermediate and final health outcomes, we retrieved 1126 unique citations. An additional 16 references were obtained from other sources. After a review of the titles and abstracts, 113 were deemed eligible for further review, and the full articles were retrieved. A total of 53 articles were found to match our inclusion criteria.8,48–99 Three studies81,91,98 reported on the same population as another included publication, and they were included as they provided additional data. Therefore, a total of 50 unique studies were included in our evaluation.

When we conducted the literature search for cancer with rhGH therapy, expanded to include GHD and ISS, 159 unique citations were retrieved and another two citations were identified through other sources. (Figure 4) One hundred-sixteen citations were excluded during the title and abstract review and 44 from the full text review. Three articles were included.100–102

Key Question 1. In patients with CF, does treatment with rhGH as an adjuvant to usual care improve intermediate outcomes, including: pulmonary function; growth (height, weight, lean body mass, protein turnover); exercise tolerance; and bone mineralization, compared with usual care alone?

Key Points

  • Ten controlled trials and eight single-arm observational studies were included.
    • Data derived from prepubertal and pubertal patients who have poor growth indices and may not be able to be extrapolated to normal growing children or adolescents.
  • Five markers of pulmonary function were evaluated in patients with CF receiving rhGH therapy.
    • In controlled trials, the FVC and percent predicted FVC significantly increased from baseline in trials comparing patients with CF receiving chronic rhGH therapy to control therapy. Single-arm observational studies support these findings.
    • In controlled trials, the FEV1 significantly increased from baseline in patients with CF receiving chronic rhGH therapy versus control therapy but the percent predicted FEV1 was not significantly improved versus control. Single-arm observational studies support the FEV1 findings but the percent predicted FEV1 findings are mixed.
    • In the one available controlled trial, no change in FEV1 Z-score occurred in patients receiving rhGH for CF versus placebo therapy and no observational studies evaluated this parameter.
  • In controlled trials suitable for pooling, significant improvements in height were observed for patients with CF receiving rhGH therapy versus control therapy as measured by the change in height, height velocity, height Z-score, and height percentile. Observational studies or other trials not suitable for pooling support these findings.
  • In controlled trials, significant improvements in weight were observed for patients with CF receiving rhGH therapy versus control therapy as measured by change in weight, weight velocity, BMI, percent IBW, LBM, and weight percentile. Patients receiving rhGH therapy did not have significantly different weight Z-score or BMI Z-score than those receiving control therapy. Observational studies evaluating change in weight, weight velocity, and weight Z-score were generally supportive of improvements associated with rhGH therapy, although one crossover trial not amenable for pooling did not show any improvement in LBM in patients receiving rhGH and who received glutamine therapy.
  • Four markers of protein turnover were evaluated in patients with CF receiving rhGH therapy. In controlled trials, rhGH therapy significantly improved two markers of protein turnover (LeuOx and NOLD) and did not significantly improve LeuRa concentrations. In one observational trial, nitrogen balance was qualitatively impacted but protein synthesis was unchanged.
  • In controlled trials, rhGH therapy significantly improved exercise work rate. Other measures of exercise tolerance were sparsely reported, so the impact is difficult to determine at this time.
  • In controlled trials and observational studies, treating patients with rhGH therapy does not improve bone age in patients with CF. However, bone mineral content does significantly improve with rhGH therapy in trials, and bone mineral content Z-score was also improved in the one trial in which it was assessed.
  • rhGH therapy in patients with CF does not seem to improve sexual maturation in males and the impact in females cannot be determined at this time. Controlled trials were not amenable to pooling and no observational trial data was available. In five controlled trials, rhGH therapy did not improve sexual maturation regardless of gender. In one controlled trial, mean Tanner stage regardless of gender improved and in an analysis of three controlled trials, rhGH therapy significantly improved sexual maturation in females but not in males.

Detailed Analysis

Study Design and Population Characteristics

Controlled Trials. Eighteen publications of controlled trials, which represent ten unique trials (n=312), met inclusion criteria (Table 3Table 5)4,16,24–47 Three of the identified publications were abstracts,29,37,39 two of which were trials published as full articles (also identified in the search),29,37 and one which has not yet been published as a full article.39

Table 3. Study design and population of controlled trials which evaluated rhGH.

Table 3

Study design and population of controlled trials which evaluated rhGH.

Table 4. Baseline characteristics of patients in controlled trials evaluating rhGH.

Table 4

Baseline characteristics of patients in controlled trials evaluating rhGH.

Table 5. Baseline characteristics of patients in controlled trials evaluating rhGH.

Table 5

Baseline characteristics of patients in controlled trials evaluating rhGH.

Four publications reported results on the same patients as another published trial.28,31,36,38 One publication31 was an interim analysis, so only results from the latter publication30 are used in our CER. Another publication was a substudy looking at a single site36 within a multi-center trial.4 Another publication38 reports new data on sexual maturation by pubertal status and gender on patients who were enrolled in three aforementioned prospective trials.24,25,34,103

Of the 10 trials, 8 trials compared rhGH to no treatment,16,24–27,33–35 1 trial used a placebo control,4 and 1 trial compared rhGH alone to either glutamine or the combination of glutamine and rhGH.30 Two trials used a crossover design,26,30 while the others used a parallel study design.4,16,24,25,27,33–35 Only one trial was double-blinded.4 Four trials received funding from foundations or government,16,24–26 eight trials received funding from industry,4,16,24,25,27,30,33,35 and two trials did not report a funding source.34,39 Four of the aforementioned trials received both industry and foundation funding to conduct their studies.24–26,30,33

One of the ten trials treated patients with rhGH for 4 weeks,30 while the other trials treated patients for 6 months to 1 year.4,16,24–27,33–35 Chronological age of patients was up to 23 years, but six trials specifically evaluated prepubertal children16,24–26,30,33–35 and one study evaluated only pubertal adolescents.34 Doses of rhGH ranged from 0.23 to 0.49 mg/kg/week, with the typical dose being 0.3 mg/kg/week.16,24,25,33–35 One trial evaluated two doses of rhGH compared to placebo.4 Males constituted at least half of the patients in trials, ranging from 50 to 83 percent of the total number of subjects.

Single-Arm Observational Studies. Eight observational reports (n=58), not constituted by patients in clinical trials, evaluated the use of rhGH in patients with CF (Table 6Table 7). Three were case reports,40,46,47 of which, one was in a patient with growth hormone deficiency and short stature,40 one was in a patient who had previously undergone lung transplantation,46 and another was in two patients with CF-related liver dysfunction.47 None of the studies included a comparator group. One study was funded by a foundation grant,40 four studies were funded by industry,41–43,45 and three did not report sources of funding.44,46,47

Table 6. Study design and population of single-arm observational studies evaluating rhGH.

Table 6

Study design and population of single-arm observational studies evaluating rhGH.

Table 7. Baseline characteristics of patients in single-arm observational studies evaluating rhGH.

Table 7

Baseline characteristics of patients in single-arm observational studies evaluating rhGH.

The duration of treatment with rhGH ranged from 6 months to 3 years. Ages of patients in the studies ranged from 6 months to 13 years,40–45,47 with the exception of one case report in a patient aged 18 years.46 Doses of rhGH in the studies ranged from 0.16–0.35 mg/kg/wk,41,42,44,45,47 with the exception of one case report where the dose was 2.2 mg/day.46 Two studies did not report the dose of rhGH.40,43 Baseline measures of height and weight were inconsistently reported among observational studies, but all patients had deficient height and weight for age.40–47

Outcome Evaluations

Pulmonary Function. Seven trials, summarized in Table 8, reported information on various pulmonary measures in CF patients treated with rhGH, including absolute FVC, percent predicted FVC, absolute FEV1, percent predicted FEV1, and FEV1 Z-score.4,16,24,26,27,34,35 Three observational studies also provided insight on the effect of rhGH on pulmonary function.41,42,45

Table 8. Change from baseline in pulmonary outcomes in controlled trials evaluating rhGH.

Table 8

Change from baseline in pulmonary outcomes in controlled trials evaluating rhGH.

Three trials reported the change from baseline in absolute FVC, which was amenable to quantitative synthesis.16,34,35 Upon statistical pooling, patients treated with rhGH had significantly greater improvements in absolute FVC than those without treatment (WMD 0.67 L, 95 percent CI 0.24 to 1.09 L). (Figure 5) A moderate degree of statistical heterogeneity was detected (I2=55 percent), though all three studies exhibited the same direction of effect. The individual point estimates for two trials34,35 were similar and greater in magnitude than the last trial.16 Since the primary investigator was the same for all three trials and the populations, doses, and durations of therapy were similar, there is not a ready explanation for this heterogeneity. However, the 95 percent CIs for the three trials overlapped and the trial with the greatest weight in the meta-analysis had the smallest magnitude of FVC improvement. Publication bias could not be evaluated due to an insufficient number of studies. In a single-arm, observational study (n=9), there was a nonsignificant increase in absolute FVC over 12 months of rhGH therapy (baseline 1.33±0.32 L; 12 months 1.46±0.49 L, p-value not reported).42

Figure 5. KQ1 pulmonary function—meta-analysis of change from baseline in absolute FVC in CF patients treated with rhGH.

Figure 5

KQ1 pulmonary function—meta-analysis of change from baseline in absolute FVC in CF patients treated with rhGH. Legend: CF=cystic fibrosis; FVC=forced vital capacity; rhGH=recombinant human growth hormone Note: The squares represent individual (more...)

Five trials, including a trial with two active rhGH treatment arms, reported percent predicted FVC consistently, allowing for quantitative synthesis.4,24,27,34,35 Upon statistical pooling, patients treated with rhGH experienced greater improvements from baseline in percent predicted FVC than patients in the control group (WMD 9.34 percent, 95 percent CI 3.41 to 15.27 percent). (Figure 6) A moderate degree of statistical heterogeneity was detected (I2=69.2 percent) but all studies exhibited the same direction of effect. The individual point estimates for two trials24,34 were greater in magnitude than the other, with one trial’s24 point estimate falling outside the confidence interval for the pooled effect size. However, the doses used and the duration of followup for these two trials were similar to the others, so an explanation for the heterogeneity is unclear. Publication bias was not detected in this analysis. Both dosing arms of the trial by Schnabel and colleagues showed a similar direction and magnitude of effect, suggesting a lack of dose-response relationship with rhGH therapy. In a single-arm, observational study (n=9), 12 months of rhGH therapy resulted in nonsignificant decreases in percent predicted FVC (baseline 85.6±17.9 percent; 12 months 80.7±19.7 percent, p-value not reported).42 In another single-arm observational study of nine patients over 12 months, the percent predicted FVC improved in seven patients and remained stable in 2 patients, but quantifiable data was not reported.45

Figure 6. KQ1 pulmonary function—meta-analysis of change from baseline in percent predicted FVC in CF patients treated with rhGH.

Figure 6

KQ1 pulmonary function—meta-analysis of change from baseline in percent predicted FVC in CF patients treated with rhGH. Legend: CF=cystic fibrosis; %FVC=percent predicted forced vital capacity; rhGH=recombinant human growth hormone Note: The squares (more...)

Four trials reported the change from baseline in absolute FEV1 suitable for quantitative synthesis.16,26,34,35 Statistical pooling showed patients treated with rhGH experiencing significantly greater improvements in absolute FEV1 versus control (WMD 0.23 L, 95 percent CI 0.01 to 0.46 L). (Figure 7) A moderate degree of statistical heterogeneity was detected (I2=43.2 percent) and all trials showed similar direction of effect. Assessment for publication bias was nonsignificant. In one single-arm, observational study (n=9), 12 months of rhGH therapy resulted in nonsignificant increases in absolute FEV1 (baseline 1.16±0.3 L; 12 months 1.20±0.52 L, p-value not reported).42

Figure 7. KQ1 pulmonary function—meta-analysis of change from baseline in absolute FEV1 in CF patients treated with rhGH.

Figure 7

KQ1 pulmonary function—meta-analysis of change from baseline in absolute FEV1 in CF patients treated with rhGH. Legend: CF=cystic fibrosis; FEV1=forced expiratory volume in one second; rhGH=recombinant human growth hormone Note: The squares represent (more...)

Four trials reported the change in percent predicted FEV1 from baseline and were amenable to quantitative synthesis.4,24,27,35 Upon statistical pooling, there was a nonsignificant greater improvement in the rhGH group compared to control (WMD 2.43 percent, 95 percent CI −3.99 to 8.85 percent). (Figure 8) No statistical heterogeneity or publication bias was detected. Three single arm observational studies also evaluated the impact of rhGH therapy on percent predicted FEV1. In the first study evaluating patients over the age of 6 years, five patients had baseline and 6 months data and four patients had baseline and 12 month data. The percent predicted FEV1 after 6 months of rhGH therapy was not significantly changed (baseline 74±2.17 percent; 6 months 70±11.41, p=0.43) and was also not significantly changed at 12 months (baseline 75±1.41 percent; 12 months 81.5±21.76 percent, p=0.59).41 In the second study (n=9), there was a no significant change in percent predicted FEV1 after 12 months of rhGH therapy (baseline 83.0±25.0 percent; 12 months 71.9±25.2 percent, p-value not reported).42 In the final study, the percent predicted FEV1 was reported as “improved” in seven of nine patients and remained stable in two of nine patients, but quantifiable data was not reported.45

Figure 8. KQ1 pulmonary function—meta-analysis of change from baseline in percent predicted FEV1 in CF patients treated with rhGH.

Figure 8

KQ1 pulmonary function—meta-analysis of change from baseline in percent predicted FEV1 in CF patients treated with rhGH. Legend: CF=cystic fibrosis; %FEV1=percent predicted forced expiratory volume in one second; rhGH=recombinant human growth (more...)

Only one trial reported the change in FEV1 Z-score from baseline.4 Upon statistical pooling of the lower and higher dose rhGH arms versus placebo, there was no significant effect on FEV1 Z-score (WMD −0.005, 95 percent CI −0.22 to 0.21).(Figure 9) There were too few trials to conduct evaluations of statistical heterogeneity and publication bias. Both dosing arms of rhGH showed similar null effects on FEV1 Z-score.

Figure 9. KQ1 pulmonary function—meta-analysis of change from baseline in FEV1 Z-score in CF patients treated with rhGH.

Figure 9

KQ1 pulmonary function—meta-analysis of change from baseline in FEV1 Z-score in CF patients treated with rhGH. Legend: CF=cystic fibrosis; FEV1=forced expiratory volume in one second; rhGH=recombinant human growth hormone Note: The squares represent (more...)

Anthropometrics

Height. Seven trials reported the effect of rhGH on height-related outcomes in patients with CF, including absolute height, height velocity, height Z-score, and height percentile.4,16,24,33–35,39 (Table 9) Six observational studied also reported on the effect of rhGH on height-related outcomes.40–45

Table 9. Change from baseline in height outcomes in controlled trials evaluating rhGH.

Table 9

Change from baseline in height outcomes in controlled trials evaluating rhGH.

Three trials reported on the change from baseline in height, permitting quantitative synthesis.24,26,34 Upon statistical pooling, there was significant improvement in height from baseline in the rhGH group compared to control (WMD 3.13 cm, 95 percent CI 0.88 to 5.38 cm). (Figure 10) A high degree of statistical heterogeneity was detected (I2=77.3 percent), but all studies showed the same direction of effect. The individual point estimate for one trial26 was lesser than the other two trials24,34 which were conducted by the same investigator. Although the doses of rhGH used in all three trials were similar, the trial by Hutler and colleagues was only conducted for 6 months.26 There were too few studies to assess the presence of publication bias. In a single-arm observational study of 24 patients, therapy with rhGH yielded sustained increases in height over several years of treatment, but quantifiable data was not reported.43

Figure 10. KQ1 anthropometrics—meta-analysis of change from baseline in height in CF patients treated with rhGH.

Figure 10

KQ1 anthropometrics—meta-analysis of change from baseline in height in CF patients treated with rhGH. Legend: CF=cystic fibrosis; rhGH=recombinant human growth hormone Note: The squares represent individual point estimates. The size of the square (more...)

Three trials reported the change from baseline in height velocity after treatment with rhGH, which was amenable to quantitative synthesis.4,16,24 Statistical pooling of these trials showed a significant improvement from baseline in height velocity in the rhGH group compared to control (WMD 3.27 cm/year, 95 percent CI 2.33 to 4.21 cm/year). (Figure 11) A moderate degree of statistical heterogeneity was detected (I2=38.2 percent), though all studies exhibited the same direction of effect. The statistical heterogeneity was most likely related to a more profound effect in the trial by Hardin and colleagues in 200124 and the less profound effect in the lower dosing arm of the trial by Schnabel and colleagues.4 The higher dose arm from the Schnabel trial had a similar magnitude of effect as the Hardin trial, which had the greatest weight in the meta-analysis. No publication bias was detected.

Figure 11. KQ1 anthropometrics—meta-analysis of change from baseline in height velocity in CF patients treated with rhGH.

Figure 11

KQ1 anthropometrics—meta-analysis of change from baseline in height velocity in CF patients treated with rhGH. Legend: CF=cystic fibrosis; rhGH=recombinant human growth hormone Note: The squares represent individual point estimates. The size of (more...)

In a case report of a 9-year-old female with CF, height velocity increased from 3.2 cm/year to 12 cm/year, yielding a 2 cm increase in height within the first 2 months of rhGH therapy.40 In a single-arm observational study, the first 6 months of rhGH yielded increases in height velocity in all seven patients studied (range 0.33 to 4.14 cm/year), with four patients experiencing clinically significant increases (defined as a greater than 2 cm/year increase).41 In another single-arm observational trial (n=9), height velocity significantly increased after 12 months of rhGH therapy, from 5.7±0.2 cm/year before therapy to 7.8±0.4 cm/year after therapy (p<0.05).42 Upon discontinuation of rhGH, height velocity declined to 4.5±0.6 cm/year in the year without rhGH therapy and was significantly lower than the year prior to therapy (p<0.05).42 Height velocity also significantly increased in 7 patients out of the 24 evaluated in another single-arm evaluation after 1 year of rhGH therapy (p<0.05) but other data was not provided.43 The last single-arm observational study (n=9) showed significant increases in height velocity from baseline over 12 months of rhGH treatment (p=0.01) but other data was not provided.45

Three trials reported the change from baseline in height Z-score, allowing for quantitative synthesis.24,35,39 Statistical pooling resulted in a significantly greater improvement from baseline in rhGH-treated patients than control (WMD 0.51, 95 percent CI 0.35 to 0.66).(Figure 12) No statistical heterogeneity was detected. Publication bias could not be evaluated because there were too few trials. One additional trial did not report the change from baseline and could not be included in the quantitative synthesis, but found that at the end of 12 months of treatment, there was no significant difference in height Z-score between the rhGH group and the control group (−1.09±0.8 versus −1.99±0.89, p-value not reported).33 In a single-arm observational trial (n=9), height Z-score significantly improved from −1.3±0.23 to −0.76±0.23 (p<0.05) after 12 months of rhGH treatment.42 Discontinuation of rhGH resulted in height Z-scores returning to pretreatment values during the year after therapy, but quantifiable data was not reported.42 Another single-arm evaluation (n=5) found that height Z-score significantly improved after 12 months of rhGH therapy (baseline −2.80±0.60; 12 months −1.56±0.60, p<0.01).44 There was also a significant improvement at 24 months of therapy (−0.94±0.40, p<0.02 versus baseline and versus 12 months).44 In another nine patients treated with rhGH in a single-arm evaluation, the height Z-score also significantly improved from baseline (baseline−1.86±0.7; 12 months −1.31±0.9, p=0.03).45

Figure 12. KQ1 anthropometrics—meta-analysis of change from baseline in height Z-score in CF patients treated with rhGH.

Figure 12

KQ1 anthropometrics—meta-analysis of change from baseline in height Z-score in CF patients treated with rhGH. Legend: CF=cystic fibrosis; rhGH=recombinant human growth hormone Note: The squares represent individual point estimates. The size of (more...)

One trial reported on the effect of rhGH on height percentile.24 After 12 months of therapy, the rhGH group experienced significant improvement from baseline in height percentile (baseline 7.5±1.2; 12 months 20.0±1.4, p=0.032).24 Changes in the control group were not significant (baseline not reported; 12 months 7.8±1.6, p=0.64).24

Weight. Ten trials reported outcomes on weight in CF patients treated with rhGH, including absolute weight, weight velocity, weight Z-score, weight percentile, body mass index (BMI), BMI Z-score, percent of ideal body weight, and lean body mass (LBM).4,16,24,26,27,30,33–35,39 (Table 10) Four observational studies also reported results on weight outcomes.41,42,44,45

Table 10. Change from baseline in weight outcomes in controlled trials evaluating rhGH.

Table 10

Change from baseline in weight outcomes in controlled trials evaluating rhGH.

Five trials reported the change from baseline in body weight, permitting quantitative synthesis.4,24,26,34,39 Upon statistical pooling, there was significantly greater improvement in the rhGH group than the control group from baseline (WMD 1.48 kg, 95 percent CI 0.62 to 2.33 kg). (Figure 13) A moderate degree of statistical heterogeneity was detected (I2=49 percent), but all studies showed the same direction of effect, and all but one34 showed a similar magnitude of effect. The trial by Hardin and colleagues in 2005 evaluated adolescent patients exclusively,34 who may have shown different responses to rhGH than the remaining trials. No publication bias was detected. The two dosing arms of the study by Schnabel and colleagues showed similar magnitude and direction of effect, suggesting a lack of dose-response relationship on body weight. In one single-arm observational trial (n=9), body weight was significantly increased after 1 year of rhGH therapy (baseline 21.5±3.1 kg, 12 months 24.9±4.2, p=0.007).45

Figure 13. KQ1 anthropometrics—meta-analysis of change from baseline in weight in CF patients treated with rhGH.

Figure 13

KQ1 anthropometrics—meta-analysis of change from baseline in weight in CF patients treated with rhGH. Legend: CF=cystic fibrosis; rhGH=recombinant human growth hormone Note: The squares represent individual point estimates. The size of the square (more...)

Two trials reported the change in weight velocity from baseline, allowing for quantitative synthesis.16,24 Upon statistical pooling, the rhGH group showed significantly greater improvements in weight velocity from baseline compared to control (WMD 2.15 kg/year, 95 percent CI 1.52 to 2.78 kg/year). (Figure 14) There were too few studies to evaluate for statistical heterogeneity or publication bias, but both studies showed similar direction and magnitude of effect. In a single-arm observational trial evaluating rhGH, the weight velocity did not significantly change from baseline in seven patients treated for 6 months (baseline 1.84±2.52 kg/year; 6 months 3.15±1.69, p=0.24).41 The p-value derived from comparing the 6 month time period to baseline was reported by the authors as being 0.03, but statistical analysis by our group using the raw data yielded a p-value of 0.24 using a paired t-test (Primer of Biostatistics: The Program, Dubeque, IA). Only four patients were treated for 12 months and their weight velocity showed no change compared to baseline (baseline 2.81±1.03 kg/year; 12 months 6.58±3.46 kg/year, p=0.07).41 Another single-arm evaluation reported that weight velocity did not significantly change during or after rhGH therapy (quantifiable data and p-value not reported).42 After 12 months of rhGH therapy in a third observational study (n=9), weight velocity significantly improved from baseline (1.7±0.9 to 3.8±1.6 kg/year, p=0.03).45

Figure 14. KQ1 Anthropometrics - Meta-analysis of change from baseline in weight velocity in CF patients treated with rhGH.

Figure 14

KQ1 Anthropometrics - Meta-analysis of change from baseline in weight velocity in CF patients treated with rhGH. Legend: CF=cystic fibrosis; rhGH=recombinant human growth hormone Note: The squares represent individual point estimates. The size of the (more...)

Four trials reported the change in weight Z-score from baseline, which was amenable to quantitative synthesis.24,27,33,35 Upon statistical pooling, there was no statistical improvements in weight Z-score in the rhGH group compared to control (WMD 0.49, 95 percent CI −0.02 to 1.00). (Figure 15) A moderate degree of statistical heterogeneity was detected (I2=63.8 percent), though all but one trial27 showed similar direction and magnitude of effect. The dose of rhGH and the duration evaluated in the trial by Schibler and colleagues27 was the same as what was studied by Hardin and colleagues,24,33,35 so this heterogeneity is not readily explained. No significant publication bias was noted.

Figure 15. KQ1 anthropometrics—meta-analysis of change from baseline in weight Z-score in CF patients treated with rhGH.

Figure 15

KQ1 anthropometrics—meta-analysis of change from baseline in weight Z-score in CF patients treated with rhGH. Legend: CF=cystic fibrosis; rhGH=recombinant human growth hormone Note: The squares represent individual point estimates. The size of (more...)

In one single-arm observational study, weight Z-score significantly improved after 12 months of rhGH therapy (baseline −1.95±0.51; 12 months −0.97±0.56, p<0.01).44 Weight Z-score was additionally improved at 24 months of therapy (−0.11±0.11, p<0.02 versus baseline and versus 12 months).44

One trial reported on the effect of rhGH on weight percentile.24 After 12 months of therapy, the rhGH group experienced a significant improvement in weight percentile (baseline 4.0±1.5; 12 months 9.0±1.3, p=0.042).24 There were no significant changes in the control group (baseline not reported; 12 months 3.5±1.9, p-value not reported).24

Two trials reported the change from baseline in BMI, permitting quantitative synthesis.34,35 Statistical pooling resulted in significantly greater improvements from baseline in BMI in the rhGH group compared to control (WMD 2.08 kg/m2, 95 percent CI 1.20 to 2.96 kg/m2). (Figure 16) There were too few studies to evaluate statistical heterogeneity or publication bias, but both studies showed similar direction and magnitude of effect. One single-arm observational study (n=7) reported that there was no significant change in BMI from baseline, but quantifiable data was not reported.41

Figure 16. KQ1 anthropometrics—meta-analysis of change from baseline in BMI in CF patients treated with rhGH.

Figure 16

KQ1 anthropometrics—meta-analysis of change from baseline in BMI in CF patients treated with rhGH. Legend: BMI=body mass index; CF=cystic fibrosis; rhGH=recombinant human growth hormone Note: The squares represent individual point estimates. The (more...)

One trial reported the effect of rhGH on the change from baseline in BMI Z-score.4 This trial did not evaluate change from baseline in BMI like the aforementioned trials.34,35 This trial evaluated two dosing arms of rhGH and was amenable to quantitative synthesis. After pooling the two dosing arms of the trial, there was no significant difference between the rhGH group and placebo group on BMI Z-score (WMD −0.05, 95 percent CI −0.30 to 0.20). (Figure 17) There were too few studies to evaluate statistical heterogeneity or publication bias, but both dosing arms of the trial showed similar direction and magnitude of effect. This suggests a lack of a dose-response relationship.

Figure 17. KQ1 anthropometrics—meta-analysis of change from baseline in BMI Z-score in CF patients treated with rhGH.

Figure 17

KQ1 anthropometrics—meta-analysis of change from baseline in BMI Z-score in CF patients treated with rhGH. Legend: BMI=body mass index; CF=cystic fibrosis; rhGH=recombinant human growth hormone Note: The squares represent individual point estimates. (more...)

Two trials reported the effect of rhGH therapy on change from baseline in percent of IBW, permitting quantitative synthesis.24,34 Upon statistical pooling, there was a significantly greater improvement from baseline in percent IBW in the rhGH group compared to control (WMD 12.57, 95 percent CI 7.01 to 18.12). (Figure 18) There were too few studies to evaluate statistical heterogeneity or publication bias, but both studies showed similar direction and magnitude of effect.

Figure 18. KQ1 anthropometrics—meta-analysis of change from baseline in percent IBW in CF patients treated with rhGH.

Figure 18

KQ1 anthropometrics—meta-analysis of change from baseline in percent IBW in CF patients treated with rhGH. Legend: CF=cystic fibrosis; %IBW=percent ideal body weight; rhGH=recombinant human growth hormone Note: The squares represent individual (more...)

Eight trials reported the change in LBM from baseline in patients treated with rhGH, and were amenable to quantitative synthesis.4,16,24,26,27,33,35,39 Upon statistical pooling, the rhGH group showed significantly greater improvements from baseline in LBM compared to control group (WMD 1.92 kg, 95 percent CI 1.47 to 2.37 kg). (Figure 19) A low degree of statistical heterogeneity was detected (I2=20.9 percent) and all studies showed similar direction and magnitude of effect. Publication bias was unlikely. In a crossover trial not included in quantitative synthesis that evaluated rhGH, glutamine, or the combination of the two (n=9), there was no significant difference between the treatment groups at the end of the 4 week treatment period in LBM as measured by DEXA (rhGH 22.3±5.7 kg; rhGH and glutamine combination 22.4±4.2 kg; glutamine 21.4±4.5 kg, p-value not reported).30 This trial was not included in quantitative synthesis because it evaluated rhGH either alone or in combination with glutamine versus a glutamine control, rather than a nonactive control like in the other trials.30

Figure 19. KQ1 anthropometrics—meta-analysis of change from baseline in lean body mass in CF patients treated with rhGH.

Figure 19

KQ1 anthropometrics—meta-analysis of change from baseline in lean body mass in CF patients treated with rhGH. Legend: CF=cystic fibrosis; rhGH=recombinant human growth hormone Note: The squares represent individual point estimates. The size of (more...)

Protein Markers. Two trials reported results on markers of protein catabolism subsequent to treatment with rhGH.25,30 (Table 11) One observational study also reported the effect of rhGH on markers of protein catabolism.42

Table 11. Change from baseline in protein markers in controlled trials evaluating rhGH.

Table 11

Change from baseline in protein markers in controlled trials evaluating rhGH.

In a parallel, randomized controlled trial, there was no significant change in leucine rate of appearance (LeuRa) from baseline in either the rhGH (−44±15 μmol/kg/hr) or control group (10±28 μmol/kg/hr) after 12 months of treatment, although qualitative improvements were seen.25 Treatment with rhGH for 12 months significantly improved LeuOx, NOLD, and LeuOx/NOLD ratio (change from baseline: −19±7 μmol/kg/hr, −30±16 μmol/kg/hr, and −0.06±0.02, respectively, p<0.05 for all comparisons to baseline).25 The control group experienced no significant changes from baseline in any of these parameters.25 (Table 11) Data was not reported in a manner that we could calculate the intergroup p-values.

In a crossover trial, Darmaun and colleagues evaluated the effects of rhGH, glutamine, or their combination on LeuRa, LeuOx, and NOLD.30 The LeuRa concentration at baseline was 2.89±0.22 μmol/kg of LBM/min.30 No significant changes in LeuRa resulted after treatment with glutamine alone (2.82±0.18 μmol/kg of LBM/min, p=0.48 versus baseline), rhGH alone (2.96±0.27 μmol/kg of LBM/min, p-value not reported), or rhGH and glutamine combination (2.98±0.30 μmol/kg of LBM/min, p=0.69).30 Treatment with either rhGH alone (0.49±0.09 μmol/kg of LBM/min versus 0.72±0.05 μmol/kg of LBM/min, p=0.004) or the rhGH plus glutamine combination 0.46±0.08 μmol/kg of LBM/min versus 0.70±0.05 μmol/kg of LBM/min, p=0.01) significantly improved LeuOx versus baseline, while there was no significant change for glutamine alone (0.64±0.10 μmol/kg of LBM/min versus 0.71±0.05 μmol/kg of LBM/min, p=0.36).30 For the NOLD endpoint, the rhGH alone (2.41±0.22 μmol/kg of LBM/min versus 2.13±0.22 μmol/kg of LBM/min, p=0.01) and the rhGH plus glutamine combination (2.52±0.21 μmol/kg of LBM/min versus 2.13±0.22 μmol/kg of LBM/min, p=0.05) groups showed statistically significant improvements from baseline, while the glutamine alone group remained unchanged (2.18±0.18 μmol/kg of LBM/min versus 2.18±0.22 μmol/kg of LBM/min, p-value not reported).30

In one single-arm observational study (n=9) which evaluated markers of protein metabolism, nitrogen balance was negative in all patients prior to beginning rhGH, but became less negative in five patients after treatment.42 After 12 months of rhGH therapy, protein turnover changed from 5.6±0.5 g/kg/day before treatment to 5.2±0.5 g/kg/day (p-value not reported).42 Protein synthesis remained unchanged over 12 months of rhGH therapy (3.9±0.3 g/kg/day before treatment to 3.9±0.4 g/kg/day at 12 months, p-value not reported).42 In patients who achieved positive net protein anabolism (n=5), net protein anabolism changed from −0.6±0.1 g/kg/day before treatment to 0.317±0.07 g/kg/day at 12 months (p-value not reported).42

Exercise Tolerance. Three randomized controlled trials4,26,27 evaluated exercise tolerance, all using a bicycle ergometer test, and reporting endpoints including exercise work rate, oxygen consumption, maximal oxygen consumption, oxygen pulse, and peak ventilation rate. (Table 12)

Table 12. Change from baseline in exercise tolerance in controlled trials evaluating rhGH.

Table 12

Change from baseline in exercise tolerance in controlled trials evaluating rhGH.

Two trials reported the change from baseline in exercise work rate, allowing for quantitative analysis.4,27 Upon statistical pooling, the exercise work rate there was no statistically significant difference between the rhGH group compared to control (WMD 11.80 W, 95 percent CI −0.44 to 24.04 W). (Figure 20) A low degree of statistical heterogeneity was detected (I2=23.7 percent), but all studies exhibited the same direction of effect. Publication bias could not be evaluated because there were too few studies. The two dosing arms of the trial by Schnabel and colleagues showed similar direction and magnitude of effect, suggesting a lack of dose-response relationship with rhGH therapy.

Figure 20. KQ1 exercise tolerance — meta-analysis of change from baseline in exercise work rate in CF patients treated with rhGH.

Figure 20

KQ1 exercise tolerance — meta-analysis of change from baseline in exercise work rate in CF patients treated with rhGH. Legend: CF=cystic fibrosis; rhGH=recombinant human growth hormone Note: The squares represent individual point estimates. The (more...)

The remaining endpoints were sparsely reported and thus not amendable to quantitative analysis. The trial by Hutler and colleagues was a crossover trial comparing rhGH therapy to control, and their data was reported as both separate time periods and as combined treatment groups.26 When looking at only the first period of data, there appeared to be a greater improvement from baseline in those treated with rhGH in peak oxygen uptake (VO2-peak) (change from baseline in rhGH 201±161 mL versus control −18±117 mL, p-value not reported).26 There were also improvements from baseline in the rhGH group during the first period of treatment compared to control in oxygen pulse peak (rhGH 1.0±0.7 ml/beat versus control −0.1±0.5 ml/beat, p-value not reported) and in ventilation peak (rhGH 5.3±6.6 versus control −0.4±5.5, p-value not reported).26 From the crossover data in which data from the same treatment groups were combined, there were significantly greater improvements from baseline in rhGH group compared to control in exercise power (p=0.008), VO2 (p=0.009), and oxygen pulse (p=0.008).26 In the study by Schibler and colleagues, the maximal oxygen consumption (VO2max) remained unchanged in the rhGH group (baseline 40.7±2.7 ml/kg/min; 12 months 38.2±2.1 ml/kg/min, p-value not reported) but the control group showed significant decreases in VO2max (baseline 44.1±3.5 ml/kg/min; 12 months 35.5±2.5 ml/kg/min, p=0.003).27 Schnabel and colleagues found that maximal oxygen consumption (in ml/min) increased from baseline in both doses of rhGH treatment groups and that this change was significantly greater than the change from baseline in the placebo group (p<0.05 for both dose groups versus placebo).4 During the open-label treatment of all patients following the double-blind study, patients originally treated with placebo showed improvements in work rate (6.1±16.6 W) and maximal oxygen consumption (86.9±220.4 ml/min) after being treated with rhGH.4

One single-arm observational study reported a decline in exercise endurance time in all five patients studied during the first 6 months of treatment (p-value not reported), but this resolved in the four patients who completed the study at 12 months (p-value not reported).41

Bone Mineralization. Five trials reported bone mineralization outcomes in CF patients being treated with rhGH, including bone age, bone mineral content, and bone mineral content Z-score.16,24,33–35 (Table 13) Three observational studies also reported on changes in bone age subsequent to rhGH therapy.41,42,44

Table 13. Change from baseline in bone mineralization outcomes in controlled trials evaluating rhGH.

Table 13

Change from baseline in bone mineralization outcomes in controlled trials evaluating rhGH.

Two trials reported change from baseline in bone age.24,33 While this data is amenable to quantitative synthesis, the clinical implication of the value attained from statistical pooling is uncertain. Therefore, this endpoint is reported qualitatively. After 12 months of treatment, Hardin and colleagues reported that the change in bone age from baseline in the rhGH group was 1.1±0.9 years and the control group was 0.9±1.2 years.24 The p-value was reported as nonsignificant, but it is unclear if it refers to the comparison from the end of 12 months to baseline, or the comparison between treatment groups.24 In adolescent patients with CF studied by Hardin and colleagues, bone age was similar between groups at baseline (rhGH 14.4±1.9 years versus control 14.1±1.2 year, p>0.05) and there was no significant differences in bone age after 1 year of treatment (rhGH 15.2±1.9 year versus control 14.9±0.9 years, p=0.7).34

In a single-arm observational study (n=7), bone age advanced faster than chronological age in four patients. The mean change in bone age in all patients was 0.79±0.42 years, but this was not significantly different from the change in chronological age (0.62±0.10 years, p=0.28).41 Another single-arm observational study showed changes in bone age from baseline to be similar to change in chronological age (1.0±0.3 years over 12 months of rhGH therapy).42 A third single-arm observational study found that bone age was not significantly improved over the first 12 months of therapy (baseline 2.0±1.0 versus 12 months 2.9±1.05, p-value not reported), nor at 24 months of therapy (24 months 3.6±1.3, p-value not reported).44

Four trials reported change from baseline in bone mineral content, permitting quantitative synthesis.16,33–35 Pooling the data resulted in a significant improvement in bone mineral content in the rhGH group compared to control (WMD 192 g, 95 percent CI 110 to 273 g). (Figure 21) A high degree of statistical heterogeneity was detected, likely due to differences in the magnitude of effect, though the direction of effect was similar in all studies. One trial by Hardin and colleagues in 200534 exhibited a more profound effect than the other trials, possibly due to it being comprised exclusively of adolescent patients, who may have accumulated a greater bone mass due to their pubertal status and presence of sex hormones. No significant publication bias was seen in this analysis.

Figure 21. KQ1 bone mineralization — meta-analysis of change from baseline in BMC in CF patients treated with rhGH.

Figure 21

KQ1 bone mineralization — meta-analysis of change from baseline in BMC in CF patients treated with rhGH. Legend: CF=cystic fibrosis; rhGH=recombinant human growth hormone Note: The squares represent individual point estimates. The size of the (more...)

Bone mineral content (BMC) Z-score was reported in one trial16 and found to have significantly improved in patients treated with rhGH compared to control. At baseline, BMC Z-score was −2.1±0.6 in the rhGH group compared to −1.7±0.9 in the control group; at 12 months, the rhGH group had a value of −1.4±0.8 versus −1.7±0.8 in control.16 The authors provided a p-value of 0.04 at the end of this statement, but it is unclear for which comparison it refers.16 The rhGH group had a statistically significant increase in BMC Z-score from baseline compared to control (Table 13, p=0.001). In a 1 year open-label extension in which patients originally assigned to the control group received rhGH therapy, there was also an improvement in BMC Z-score up to −1.3±0.7 at the end of the study.16

Sexual Maturation. Pubertal status was reported in seven trials.16,24,26,30,33,35 (Table 14) In five trials,16,26,30,33,35 all patients were prepubertal (Tanner stage 1) and did not progress over the randomized controlled portion of the trials. In the trial by Hardin and colleagues in 2001, all patients started at Tanner stage 1; at the end of 12 months of therapy, none of the males progressed in Tanner stage, and three and two females in the rhGH and control groups progressed to Tanner stage 2, respectively.24 The trial by Hardin and colleagues in 2005 evaluated pubertal patients exclusively and reported the mean Tanner stage at baseline (rhGH 3.6±0.4; control 3.4±0.6) and study end in both groups (rhGH 4.5±0.6; control 4.1±0.9, p=not significant).34

Table 14. Outcomes of sexual maturation in controlled trials evaluating rhGH.

Table 14

Outcomes of sexual maturation in controlled trials evaluating rhGH.

One publication38 reported new data on sexual maturation by pubertal status and gender on patients who were enrolled in three prospective trials (including the Hardin 2001 trial noted above).16,24,34 More prepubertal females treated with rhGH exhibited breast development in the first 6 months than females in the control group (50 percent vs 23 percent, p<0.02).38 In prepubertal males, nonsignificant improvements in testicular development was seen in the first 6 months of rhGH treatment compared to control (25 percent vs 12.8 percent, p=0.14).38 Pubertal onset with respect to chronological age was normalized in both prepubertal females and males treated with rhGH.38 In patients who had already reached puberty before initiating rhGH, treatment did not significantly alter further pubertal development compared to control.38

Discussion

In a population with CF and impaired baseline growth indices, treatment with rhGH improved in pulmonary function as measured by absolute FVC (0.67 L improvement), percent predicted FVC (9.34 percent improvement), and absolute FEV1 (0.23 L improvement). There were no significant effects on percent predicted FEV1 and FEV1 Z-score upon statistical pooling of trials from 6 to 12 months of duration. The nonsignificant effects on percent predicted FEV1 in the face of significant improvements in absolute FEV1 are likely due to the concurrent improvements in height. Since predicted values of FEV1 are hinged upon a patient’s height,12 concurrent clinical improvements in both absolute FEV1 and height may attenuate or nullify improvements in percent predicted FEV1. It seems that pulmonary function improves with rhGH therapy but may not markedly improve above that which is caused by height improvements. The Cystic Fibrosis Foundation (CFF) pulmonary guidelines currently do not contain recommendations regarding the use of rhGH to improve pulmonary function.104 In CF patients with moderate to severe lung disease, the CFF strongly recommends the use of inhaled tobramycin and recombinant human DNase based at least partially upon the ability to improve percent predicted FEV1 by 7.8 to 12 percent (with tobramycin) and absolute FEV1 11.2 to 15.4 percent (with DNase).104 Although rhGH was unable to provide similar benefits on percent predicted FEV1, inhaled tobramycin and DNase do not affect linear growth and assert effects on pulmonary function independently.104

Most of the anthropometrics evaluated in CF patients treated with rhGH significantly improved over a treatment range of 6 to 12 months. Significant improvements in height outcomes were detected with a 3.13 cm greater height gain in rhGH-treated patients than control, and a 3.27 cm/year greater height velocity than control. Height Z-score was also improved by 0.51. Similarly, body weight was significantly improved upon statistical pooling, with improvements of 1.48 kg in patients treated with rhGH. Treatment was associated with improvements in weight velocity by 2.15 kg/year and weight Z-score by 0.49. Treatment with rhGH provided significant improvements in BMI (with a gain of 2.08 kg/m2), though BMI Z-score was not affected. Since the two studies evaluating BMI did not evaluate BMI Z-score, we cannot determine why BMI was increased in two of the studies but BMI Z-score was not impacted in another study. Percent of IBW and LBM were significantly improved by 12.57 percent and 1.92 kg, respectively. There is some evidence to suggest that low anthropometric values are associated with decreases in pulmonary function. In the Epidemiologic Study of Cystic Fibrosis, multivariate analysis showed a significant decline in percent predicted FEV1 with low weight-for-age percentile in CF patients aged 9 to 12 years (n=1696, p=0.029) and CF patients aged 13 to 17 years (n=1359, p=0.021).105 Lower values of weight-for-age and height-for-age are also associated with low levels of percent predicted FEV1 later in life, with anthropometrics ate age 3 years being correlated with pulmonary function at age 6 years.7

To provide clinical context to the absolute changes in height and weight, we calculated the height and weight percentiles and Z-scores for a hypothetical CF patient with typical characteristics. Based on unpublished data by Hardin and colleagues, we determined that a typical prepubertal CF patient at baseline is aged 9.33 years, is 125 cm tall, and weighs 25 kg. Using the WHO AnthroPlus software for a male patient who is 9.33 years old, 125 cm tall, and 25 kg heavy, this patient has a baseline height percentile of 6.4, height Z-score of −1.52, weight percentile of 15.4, and weight Z-score of −1.02. Without the use of rhGH, the patient would gain 1 kg of body weight and 0.2 cm of height in 1 year, and would have height percentile 1.4, height Z-score −2.21, and incalculable weight percentile and weight Z-score. If rhGH were administered for 1 year with additional height increases of 3.13 cm and weight gain of 1.48 kg over control, this patient at 1 year would have height percentile 4.2, height Z-score −1.72, and incalculable weight percentile and Z-score. While the rhGH-treated patient has not achieved population norms, his values are closer to normal than without rhGH therapy.

In both trials evaluating protein turnover, rhGH therapy significantly improved two protein markers (LeuOx and NOLD) but did not significantly impact LeuRa, although qualitative improvements in this marker were seen in both trials. In the observational, single arm trial, nitrogen balance was less negative but protein synthesis was unaltered. LeuRa is based on the rate of isotopically-labeled leucine release from tissues (due to protein breakdown) into the intracellular space.106 Since leucine can be oxidized in muscle tissue,106 LeuOx is measured to aid in the calculation of NOLD, which represents whole body protein synthesis.25 Although there are no standard published values that correlate with a clinically significant change, improvement in protein kinetics can be helpful due to the catabolic condition of CF.25 Given the small sample sizes, different comparators in the three studies (no treatment, glutamine treatment, and no control group), different study types (parallel trial, crossover trial, observational study), and different means of comparison, the strength of evidence for this endpoint is insufficient.

Measures of exercise tolerance were also improved with rhGH therapy, with maximal work rate improving by 11.80 W in patients treated with rhGH compared to control. Maximal work rate is measured by setting up a bicycle ergometer at an initial work rate, increasing the work rate at predetermined increments per unit of time (either with Conconi protocol or Borg scale),4,27 and halting the exam at subjective physical exhaustion. The work rate at the time of exam completion is recorded as the maximal work rate, representing the point at which a patient cannot tolerate physical activity any further. Other endpoints related to exercise tolerance were sparsely reported and thus quantitative synthesis was not performed. One study reported improvements in peak oxygen uptake, oxygen pulse peak, and ventilation peak. Another study showed no changes in maximal oxygen consumption.

Bone mineralization was another intermediate outcome of interest. After 1 year of therapy, there was no difference in bone age between rhGH-treated patients and control. Bone mineral content was significantly improved by 192 g in rhGH-treated patients versus control upon statistical pooling. In the one trial evaluating the endpoint, bone mineral content Z-score was also found to have significantly improved by 0.7 in rhGH-treated patients after 1 year of therapy. Bone mineralization deficiencies are problems in patients with CF for several underlying reasons: vitamin D malabsorption, poor nutrition, physical inactivity, or delayed pubertal development.107 Several small randomized controlled trials may support the use of bisphosphonates in CF patients, showing increases in bone mineral density (BMD) up to 5.8 percent in the lumbar spine versus control (p<0.001) with pamidronate.108 However, the bisphosphonate trials were conducted in adult patients and presented results in percent change in bone mineral density, making comparisons difficult with the mostly pediatric population studied with rhGH and the results presented as absolute change in total body BMC. Therefore, the relevance of the changes seen in BMC with rhGH therapy is unclear.

rhGH therapy does not seem to improve sexual maturation in males with CF and the impact in females with CF cannot be determined at this time. In five controlled trials, rhGH therapy did not improve sexual maturation regardless of gender. In one controlled trial, mean Tanner stage regardless of gender improved in all patients and in an analysis of three controlled trials, rhGH therapy significantly improved sexual maturation in females but not in males.

While improvements in intermediate outcomes with rhGH therapy may be beneficial to the patient, it is essential to determine the effect of rhGH on final health outcomes. Key Question 2 seeks to evaluate the effect of rhGH on final health outcomes in CF patients, while Key Question 3 seeks to elucidate the linkages between intermediate and final health outcomes.

Key Question 2. In patients with CF, does treatment with rhGH as an adjuvant to usual care improve health outcomes, including: frequency of required intravenous antibiotic treatments; frequency of hospitalization; quality of life; bone fracture or development of osteoporosis/osteopenia; or mortality, compared with usual care alone?

Key Points

  • There is insufficient evidence to determine the effect of rhGH on IV antibiotic use during therapy.
  • There is insufficient evidence to determine the effect of rhGH on pulmonary exacerbations.
  • There is moderate evidence to suggest that rhGH therapy reduces the rate of hospitalization.
  • There is insufficient evidence to determine the effect of rhGH on HRQoL in patients with CF.
  • There is insufficient evidence to determine the effect of rhGH on bone consequences or mortality.

Detailed Analysis

Study Design and Population Characteristics

Studies to answer Key Question 2 are derived from the same set of studies used to evaluate Key Question 1 and are summarized in Table 3Table 7.

Outcome Evaluations

Antibiotic Usage. Three trials, summarized in Table 15, reported information about antibiotic usage in patients with CF.24,35,103 The varying definitions of antibiotic usage precluded quantitative analysis. In the first trial, the number of outpatient IV antibiotic courses was similar between groups in the year preceding the study (rhGH group 0.9±0.7 versus control group 0.8±0.7, p-value not reported) and in the year during therapy (rhGH group 0.7±0.8 versus control group 0.9±0.7, p-value not reported).24 In patients who were receiving enteral nutrition, rhGH therapy did not affect outpatient IV antibiotic use compared to control (rhGH 0.57±0.51 versus control 0.85±0.8, units not reported, p=0.05).35 In the third trial, it was reported that no difference in IV antibiotic use occurred between the rhGH and the control groups but quantifiable data was not reported.103

Table 15. Intravenous antibiotic usage in patients in controlled trials evaluating rhGH.

Table 15

Intravenous antibiotic usage in patients in controlled trials evaluating rhGH.

Pulmonary Exacerbations. One trial reported the number of patients who experienced pulmonary exacerbations over the duration of the trial, but there was no difference between patients treated with rhGH and those treated with placebo (Table 16, p-value not reported).4

Table 16. Pulmonary exacerbations in patients in controlled trials evaluating rhGH.

Table 16

Pulmonary exacerbations in patients in controlled trials evaluating rhGH.

One single-arm observational study compared the number of pulmonary exacerbations during 6 or 12 months of rhGH therapy to the 6 to 12 months preceding rhGH therapy.41 In patients treated with rhGH for 12 months (n=4), the number of exacerbations fell from 13 to 6; in patients treated for 6 months (n=3), exacerbations fell from 10 to 4. The authors report a p-value of p=0.04 at the end of these results but it is not clear to which comparison it belongs.41

Hospitalizations. The rates of hospitalizations per year were reported consistently in four trials and were amendable to quantitative synthesis.24,34,35,103 (Table 17) Upon statistical pooling, the rate of hospitalization during the study was significantly less in those treated with rhGH than control (WMD −1.62 hospitalizations per year, 95 percent CI −1.98 to −1.26 hospitalizations per year). (Figure 22) No statistical heterogeneity or publication bias was detected upon analysis. One additional trial reported that there were no statistically significant differences in hospitalization days between treatment groups but quantifiable data was not reported.4

Table 17. Rate of hospitalizations in patients in controlled trials evaluating rhGH.

Table 17

Rate of hospitalizations in patients in controlled trials evaluating rhGH.

Figure 22. KQ2—meta-analysis of hospitalizations in CF patients treated with rhGH.

Figure 22

KQ2—meta-analysis of hospitalizations in CF patients treated with rhGH. Legend: CF=cystic fibrosis; rhGH=recombinant human growth hormone Note: The squares represent individual point estimates. The size of the square represents the weight given (more...)

Health-Related Quality of Life. Two trials reported information regarding HRQoL, using the Cystic Fibrosis Questionnaire (CFQ).4,103 (Table 18) Quantifiable data was only reported in one trial,16 precluding quantitative synthesis. Patients treated with rhGH experienced greater improvements in the weight domain than patients in the control group (change from baseline 0.4±0.8 versus 0.3±0.8, respectively, p=0.04) and in the body image domain (change from baseline 0.3±0.9 versus −0.2±0.9, p=0.03).103 No differences were seen in the remaining CFQ domains (data not reported).103 A second trial reported no major differences among treatment groups in HRQoL but quantifiable data was not reported.

Table 18. Change from baseline in health-related quality of life of patients in controlled trials evaluating rhGH.

Table 18

Change from baseline in health-related quality of life of patients in controlled trials evaluating rhGH.

Bone Consequences. Incidence of bone consequences such as development of osteoporosis, osteopenia, or fracture was not reported in trials or studies.

Mortality. Incidence of CF-related death or death from any cause was not reported in trials or studies. Through a review of the trials and studies, no apparent deaths were reported but there were patients who were lost to followup precluding firm conclusions of their dispensation.

Discussion

From the current body of evidence, the impact that rhGH therapy has on final health outcomes is difficult to quantify. Clearly, more research is needed to discern the impact of rhGH on health outcomes and trial authors need to be more forthcoming with quantifiable outcome data, even for underpowered analyses.

Upon statistical pooling of four trials, rhGH use was associated with a 1.6 fewer hospitalizations per year than those not receiving therapy. However, an additional trial reported that no significant reductions in hospitalizations occurred with rhGH therapy, but quantifiable data was not provided and the trial could not be pooled with the others. Whether the rhGH group had qualitatively fewer hospitalizations is not known.

Data on other endpoints were either sparsely or inconsistently reported, precluding quantitative analysis. One study found a 33 percent reduction in intravenous antibiotic use with borderline significance, one trial showed 22 percent nonsignificant reduction, and the third trial only provided a summary statement saying that no significant impact occurred. As such, we cannot determine the impact of rhGH on intravenous antibiotic use in CF patients.

Only one trial evaluated the impact of rhGH therapy on pulmonary exacerbations. The numbers of pulmonary exacerbations were qualitatively higher in the low and high dose rhGH groups than the placebo group with no dose response relationship seen. While one trial found significant benefits on two aspects of health related quality of life, another trial found no substantial benefits but did not quantify the data. No data was available for bone consequences of CF or mortality.

Key Question 3 seeks to elucidate the linkages between intermediate and final health outcomes but it would have been valuable to see if preliminary data in a CF population receiving rhGH would be similar.

Key Question 3. In patients with CF, what is the strength of evidence that intermediate outcomes of pulmonary function, growth, and bone mineralization are associated with improvements in the health outcomes including quality of life, bone fracture, development of osteoporosis/osteopenia or mortality?

Key Points

  • This key question evaluates the association between intermediate endpoints and final clinical outcomes in patients with CF.
  • The association between pulmonary function and mortality in patients with CF was evaluated in 28 studies.
    • Only one of three studies which evaluated FVC at baseline and mortality found a univariate association and only two of five which evaluated percent predicted FVC at baseline and mortality found a univariate association. However, only one of the aforementioned studies performed multivariate analysis and found that percent predicted FVC at baseline was a multivariate predictor. Decreases in FVC were univariate and multivariate predictors of mortality in two trials, but not in two other trials.
    • Some studies using univariate analysis found an association between measures of absolute FEV1 and mortality but other studies did not. In the only two multivariate analyses, an association was found between FEV1 and mortality in one study but no association was seen between the decline in FEV1 and mortality. The link between percent predicted FEV1 and mortality is stronger with a majority of studies finding an association between percent predicted FEV1 and mortality.
  • The association between anthropometrics and mortality in patients with CF was evaluated in 26 studies.
    • The link between height and mortality is weak with only a minority of studies reporting an association.
    • The link between different measures of weight and mortality was supported in majority studies by univariate analysis. Only one study found a multivariate relationship between weight and mortality but another multivariate analysis did not. The link between BMI and mortality is controversial with some studies showing no association, others showing only a univariate association and very few showing no multivariate association. The link between IBW and mortality was supported by several univariate associations and in the only multivariate analysis that was performed.
    • The only study evaluating the association between percent predicted weight-for-height and mortality found a multivariate association.
  • No studies evaluated the association between protein turnover and mortality.
  • The association between exercise tolerance and mortality in patients with CF was evaluated in 10 studies. The link between walk testing and mortality is weak with some studies finding no association, some finding only a univariate association and very few finding a multivariate association. The link between peak oxygen uptake during exercise testing and mortality was only supported by univariate analyses.
  • No studies evaluated the association between bone mineralization and mortality.
  • The association between pulmonary function and HRQoL in patients with CF was evaluated in 14 studies but using 10 different scales. All studies but one specified that they explored the association between percent predicted FEV1 and HRQoL but rhGH has little to no impact on this parameter (see KQ1). The last study did not specify whether the FEV1 was the absolute or percent predicted. Only four studies employed multivariate analyses (each using different questionnaires to rate HRQoL).
    • In multivariate analyses, higher percent predicted FEV1 was associated with improvements in “ways of coping” but not subjective health perception in one study, but whether this is absolute or percent predicted FEV1 is not specified. Higher percent predicted FEV1 was associated with improvements in seven of nine health domains (including social and physical functioning and chest symptoms) in another study and general well being in another study, but no association was seen between FEV1 and general health perception in the final study.
  • The association between anthropometrics and HRQoL in patients with CF was evaluated in 10 studies but using nine different scales and different anthropometric parameters. Only five studies employed multivariate analyses (each using different questionnaires to rate HRQoL).
    • In multivariate analysis, greater percent IBW was not associated with subjective health perception or coping in one study, greater BMI was only associated with improvements in body image but not any other factor including social and physical functioning and chest symptoms in another study, adequate weight gain over 2 years was associated with improvements in physical functioning but not social or emotional functioning and BMI Z-score was not associated with any of the three dimensions in one study, greater BMI was associated with lower general health perception in one study, and BMI was not associated with life satisfaction.
  • No studies evaluated the association between protein turnover and HRQoL.
  • Two studies evaluated the impact between exercise tolerance and HRQoL using two different questionnaires. Greater exercise capacity (determined by VO2peak or maximal workload) is associated with better measures of HRQoL scores in univariate analyses.
  • No studies evaluated the association between bone mineralization and HRQoL.
  • Only one study evaluated the association between pulmonary function or anthropometrics and bone consequences. In univariate analyses, there was no relationship between FEV1, FVC, or BMI and bone fracture.
  • No studies evaluated the association between protein turnover, exercise tolerance, or bone mineralization and bone consequences.

Detailed Analysis

Study Design and Population Characteristics

Thirty-four studies evaluated the relationship between intermediate outcomes and mortality.8,48–51,53–80,110,111 (Table 19) Patients in 24 studies were clinically stable.8,48–52,55–57,59–62,64,66–71,73,76,79,80 Three studies evaluated patients around the time of admission to the Intensive Care Unit.74,75,78 Seven studies included patients that were evaluated for or received lung transplantation.53,54,58,63,65,72,77 Nine studies only evaluated adult patients,53,58,59,63,72,74,75,78,112 4 studies evaluated a combination of adolescent and adult patients57,61,62,79 only 1 study evaluated a combination of children and adolescents,73 4 studies evaluated only children,48,68,76,80 and 16 studies evaluated children, adolescents, and adults.8,50–52,54–56,60,64–67,69–71,77 Seventeen studies followed patients from 1 to 25 years,50–52,56,59–61,64,66,68–71,73,75,78,79 12 studies followed patients until death or the time of analysis,8,48,49,56,57,62,65,67,72,74,76,77 4 studies followed patients until death or transplantation,53,54,58,80 and 1 study did not report the duration of followup.63

Table 19. Characteristics of studies which report on the relationship between intermediate outcomes and mortality.

Table 19

Characteristics of studies which report on the relationship between intermediate outcomes and mortality.

Fifteen studies evaluated the relationship between intermediate outcomes and HRQoL.82–93,95–97,113,114 (Table 20) Eight different generic health scales were used to rate HRQoL: Alltagsleben (Every Day Life),86 Child Health Questionnaire (CHQ),89,92 EuroQoL 5D (EQ-5D),87 Medical Outcomes Short Form 36 (SF-36),87,88 Nottingham Health Profile (NHP),84 Quality of Well-Being (QWB),83,115 Questions on Life Satisfaction,95 and the Sickness Impact Profile (SIP)85 Two CF-specific scales were also used: Cystic Fibrosis Quality of Life Questionnaire (CFQoL),90,91 and Cystic Fibrosis Questionnaire (CFQ).93,94,96–98 Descriptions of the different HRQoL measures and their interpretations are found in the Appendix Glossary.

Table 20. Characteristics of studies which report on the relationship between intermediate outcomes and HRQoL.

Table 20

Characteristics of studies which report on the relationship between intermediate outcomes and HRQoL.

Patients in all studies were clinically stable.82–93,95–98,113 Six studies only evaluated adult patients,84,85,87,90,91,96–98 three studies evaluated a combination of adolescent and adult patients,86,93,95 two studies only evaluated adolescents,88,89 two studies evaluated a combination of children and adolescents,83,92 one study only evaluated children,92 and one study evaluated children, adolescents, and adults.82 One study followed patients for up to 18 months,95 one study was a cross-sectional survey with a 1 year followup survey,87 and the remaining studies were all cross-sectional at a single timepoint.82–86,88–93,96–98,113

One study evaluated the relationship between intermediate outcomes and bone consequences.99 (Table 21) This was a retrospective cohort study which evaluated adult patients referred for lung transplantation between January 1994 and December 1996.99 Patients were assessed retrospectively for the incidence of bone fracture.99

Table 21. Characteristics of studies which report on the relationship between intermediate outcomes and bone fracture.

Table 21

Characteristics of studies which report on the relationship between intermediate outcomes and bone fracture.

Outcome Evaluations

Mortality
Pulmonary Function

Twenty-eight studies evaluated the relationship between mortality and various measures of pulmonary function.49–57,59–67,69,71,72,74–80 (Appendix Table F1)

Forced Vital Capacity (FVC) at Baseline

Three studies evaluated the relationship between forced vital capacity at baseline and mortality using univariate but not multivariate analyses.54,63,65 In two of the studies, there was no significant difference in FVC at baseline between those who lived and those who subsequently died. (Ciriaco: MD 0L, 95 percent CI −0.48 to 0.48; Venuta: MD 0 L, 95 percent CI −0.58 to 0.58).54,63 In the third trial, the FVC was significantly higher in those who lived versus those who subsequently died (MD −0.27 L, p=0.006).65

Percent Predicted Forced Vital Capacity (FVC) at Baseline

Five studies evaluated the relationship between percent predicted FVC at baseline and mortality using univariate analysis,54,57,63,65,112 and of the five, only one conducted multivariate analyses.57 In two of the studies there was no significant difference in percent predicted FVC at baseline between those who lived and those who subsequently died (Ciriaco: MD −2 percent, 95 percent CI −5.35 to 9.35; Venuta: MD −2 percent, 95 percent CI −6.89 to 10.89).54,63 In the third study, the FVC was significantly higher in those who lived versus those who subsequently died (MD −4 percent, p=0.031).65 In the fourth study, no significant difference in percent predicted forced vital capacity occurred between those who survived and those who subsequently died but the effect sizes, p-values, and variance were not provided.112 In the fifth study, those who survived had a significantly greater percent predicted FVC at baseline than those who died but the effect size was not reported (p<0.001). In multivariate analysis, increasing percent predicted FVC was significantly associated with a reduction in mortality (RR 0.963, p<0.0001).57

Ten Percent Decrease in Percent Predicted Forced Vital Capacity (FVC)

Two studies evaluated the risk of death associated with a 10 percent decrease in percent predicted forced vital capacity using both univariate and multivariate analysis.51,77 In the first study, for every 10 percent decrease in percent predicted FVC, the hazard of death was significantly increased (HR 2.1, 95 percent CI 1.5 to 3.0) in univariate analysis but a 10 percent decrease in percent predicted FVC was not a multivariate predictor of mortality.77 In the second trial, for every 10 percent decrease in percent predicted forced vital capacity, the relative risk of death was significantly increased within two years in univariate (RR 1.9, 95 percent CI 1.8 to 2.1) and multivariate analysis (RR 2.0, 95 percent CI 1.8 to 2.2).51

Decline in Percent Predicted Forced Vital Capacity (FVC)

Two studies performed univariate and multivariate analysis to evaluate the relationship between mortality and decline in percent predicted forced vital capacity.56,62 In the first study, the univariate results were not reported but upon multivariate analysis, there was a significant relationship between declines in percent predicted FVC and mortality.56 In the second study, there was no significant difference in the rate of decline in percent predicted FVC per year in those who lived versus those who died in univariate (MD 0.39 percent, p=0.1) or multivariate analysis, but the effect size and measures of variance were not reported.62

Forced Expiratory Volume in One Second (FEV1) at Baseline

Six studies performed univariate, but not multivariate analysis to evaluate the relationship between FEV1 at baseline and mortality.53,54,63,65,67,72 Upon univariate analysis, one study found that the hazard of death was significantly decreased (HR 0.999, 95 percent CI 0.998 to 0.999) in those with a higher FEV1 at baseline.67 In another study, univariate analysis revealed that the risk of death was significantly decreased (RR 0.28, 95 percent CI 0.08 to 0.97) in those with a higher FEV1 at baseline.53 In four studies, patients who subsequently died had a lower FEV1 at baseline (ranging from 0.04 to 0.149 liters less) than those who lived, but FEV1 was not a significant univariate predictor of mortality in any of these studies.54,63,65,72

One study used both univariate and multivariate analysis to evaluate the relationship between FEV1 at baseline and mortality using data from the United States Cystic Fibrosis Foundation National Patient Registry in 1996.69 A statistically significant univariate relationship between FEV1 and mortality, but the effect size was not reported. Multivariate analysis revealed that each liter increase in FEV1 decreased the odds of dying (OR 0.09, 95 percent CI 0.7 to 0.11).69

Decline in Forced Expiratory Volume in One Second (FEV1)

Three studies evaluated the relationship between decline in FEV1 and mortality.71,75,76 The first study did not report the results of univariate analysis, but a significant relationship was found between decline in FEV1 and mortality upon multivariate analysis although no effect size was reported.71 The second study found that a decline in FEV1 before admission for a pulmonary exacerbation increased the hazard of death after pulmonary exacerbation (HR 0.70, 95 percent CI 0.49 to 1.00) but the results of multivariate analysis were not significant.75 The third study found an increase in the hazard of death with decline in FEV1 over the study period (HR 0.959, 95 percent CI 0.928 to 0.0991) upon univariate analysis, but did not a multivariate relationship.76

Percent Predicted Forced Expiratory Volume in One Second (FEV1) at baseline

Eleven studies50,53,57,59,61,63,65–67,72,79 evaluated the univariate relationship between percent predicted forced expiratory volume in one second at baseline, but only six evaluated the multivariate relationship.57,59,61,66,67,79 Using only univariate analysis, studies evaluating the relationship between percent predicted FEV1 at baseline and mortality had conflicting findings.50,54 Two studies, one evaluating individuals in two different clinics, found percent predicted FEV1 at baseline was significantly higher in those who lived versus those who subsequently died (Corey, Site 1: MD −40 percent, p=<0.05; Site 2: MD −40 percent, p<0.001, Ciriaco: MD −5 percent, p<0.02).50,54 In contrast, three studies found that percent predicted FEV1 at baseline was not significantly higher in those who lived versus those who subsequently died (Venuta: MD 3.4 percent, 95 percent CI −1.53 to 8.33, Vizza: MD 0 percent, p=<0.823, Stanchina: MD 4.8 percent, 95 percent CI −0.78 to 10.38).63,65,72 While another study found that there was not a significant decrease in the risk of death for individuals with a higher percent predicted FEV1 at baseline compared to those with a lower percent predicted FEV1 at baseline based on univariate analysis (RR 0.96, 95 percent CI 0.92 to 1.00).53

In studies using multivariate analyses, all six studies found a relationship between percent predicted FEV1 at baseline and mortality.57,59,61,66,67,79 Upon univariate analysis, two studies found that percent predicted FEV1 at baseline was significantly higher in those who lived compared to those who subsequently died (Moorcroft: MD −29.2 percent, p<0.001, Courtney: MD −28.3 percent, p<0.001) and for both studies percent predicted FEV1 at baseline was a significant multivariate predictor of mortality but the effect size was not provided.59,79 Three studies found a statistically significant univariate relationship between percent predicted FEV1 at baseline and mortality but did not report an effect size.57,61,66 Bell and colleagues found a statistically significant multivariate relationship but did not report effect size.61 Liou and colleagues found a decrease in the odds of dying upon multivariate analysis of percent predicted FEV1 at baseline, but did not report statistical significance (OR 0.96, NR).66 Belkin and colleagues found that there was a significant increase in the hazard of death for individuals with a percent predicted FEV1 ≤30 percent at baseline compared to those with a percent predicted FEV1 ≥30 percent in univariate (HR 3.8, 95 percent CI 2.0 to 7.5) and multivariate (HR 6.8, 95 percent CI 2.4 to19.3) analysis.77 One study found that individuals with a higher percent predicted FEV1 at baseline have a decreased hazard of death than those with a lower percent predicted FEV1 in univariate (HR 0.945, 95 percent CI 0.934, 0.956) and multivariate analysis (HR 0.953, 95 percent CI 0.931 to 0.975).67

Percent Predicted Forced Expiratory Volume in One Second (FEV1)

One study evaluated the relationship between the most recently recorded percent predicted FEV1 values recorded in the Canadian Patient Data Registry for the period of 1985–1989 and mortality using univariate and multivariate analysis.55 The study revealed a significant decrease in the hazard of death for those with a higher percent predicted FEV1 upon univariate (HR 0.93, 95 percent CI 0.92 to 0.94) and multivariate analysis (HR 0.93, 95 percent CI 0.92 to 0.94).55

Percent Predicted Forced Expiratory Volume in One Second (FEV1) Evaluated by Percent

Three studies evaluated the relationship between percent predicted FEV1 evaluated by percent and mortality using both univariate and multivariate analysis.52,60,67 The first study evaluated the relationship between mortality and percent predicted FEV1 below and above 80 percent of predicted using univariate and multivariate analysis and found that the hazard of death was significantly increased for those individuals with a percent predicted FEV1 between 60 and 80 percent when compared to those with a percent predicted FEV1 greater than 80 percent in univariate (HR 2.7, 95 percent CI 1.4 to 5.5) but not in multivariate analysis (HR 1.8, 95 percent CI 0.7 to 4.3).60 The hazard of death was significantly increased for those individuals with a percent predicted FEV1 between 40 to 59 percent when compared to those with a percent predicted FEV1 greater than 80 percent in univariate (HR 14.0, 95 percent CI 7.8 to 25.1) and multivariate analysis (HR 11.3, 95 percent CI 4.9 to 26.3).60 Finally, the hazard of death was significantly increased for those individuals with a percent predicted FEV1 below 40 percent when compared to those with a percent predicted FEV1 greater than 80 percent in univariate (HR 56.7, 95 percent CI 32.6to 98.5) and multivariate (HR 27.5, 95 percent CI 11.2 to 67.8) analysis.60

In the second study, the risk of death was significantly increased for those with a percent predicted FEV1 less than or equal to 50 percent when compared to those with a percent predicted FEV1 greater than or equal to 65 percent in univariate analysis (RR 3.7, 95 percent CI 1.8 to 7.9), but not in multivariate analysis (RR 1.1, 95 percent CI 0.4 to 2.7).52 In the third study, the risk of hazard of death was significantly higher for those with a percent predicted FEV1 less than or equal to 30 percent compared to those with a percent predicted FEV1 greater than 30 percent in univariate (HR 4.83, 95 percent CI 3.44,6.78), but not multivariate analysis.67

Percent Predicted Forced Expiratory Volume in One Second (FEV1) at the Last Recorded Visit

In one study, the relationship between the percent predicted FEV1 at the last visit and subsequent mortality was evaluated using univariate analysis.76 No relationship was seen between a one percent drop in percent predicted FEV1 and mortality (HR 0.928, 95 percent CI 0.894 to 0.968).76

Percent Predicted Forced Expiratory Volume in One Second (FEV1) Prior to Intensive Care Unit Admission

Three studies performed univariate, but not multivariate analysis, to evaluate the relationship between percent predicted FEV1 prior to admission to the Intensive Care Unit for pulmonary exacerbation.74,75,78 In the first study, the risk of death was significantly increased (RR 3.68, 95 percent CI 1.11 to 16.33) for those with a percent predicted FEV1 below 24 upon admission.74 The second study found that there was not a significant hazard (HR 1.00, 95 percent CI 0.91 to 1.02) of death for those patients with a stable percent predicted FEV1 at the time of admission.75 The third study found that there was a significant decrease in the hazard of death (HR 0.97, 95 percent CI 0.93 to 1.02) associated with higher percent predicted FEV1 values within 6 months preceding intensive care unit admission.78

Decline in Percent Predicted Forced Expiratory Volume in One Second (FEV1)

Four studies evaluated the relationship between mortality and a decline in percent predicted FEV1; four performing only univariate analyses and three performing univariate and multivariate analyses.56,62,64,78 In the first study, the univariate results were not presented but there was a significant multivariate relationship between greater rates of decline in percent predicted FEV1 beginning at age 5 years and ending at age of death but the effect size was not reported.56 The second study found that patients who died had a steeper percent predicted FEV1 decline per year than those who lived (MD 1.07 percent per year, p=0.0001) upon univariate analysis and a significant increase in the hazard of death upon multivariate analysis (HR 1.3, p=0.0001).62 The third study evaluated the univariate, but not the multivariate, relationship between the decline in percent predicted FEV1 over the 4 years preceding death and found a significant difference in percent predicted FEV1 decline per year in the 4 years preceding death (MD 6.1 percent, p<0.01) and the percent predicted FEV1 decline per year in the 2 years preceding death (MD 9.7 percent, p<0.01), however the percent predicted FEV1 decline per year between 2 and 4 years preceding death was not a significant predictor of death (MD 4.25 percent, p=0.22).64 In the final study, univariate (HR 1.25, 95 percent CI 1.04 to 1.52) and multivariate (HR 1.47, 95 percent CI 1.18 to 1.85) analysis revealed that a decline in percent predicted FEV1 per year significantly increased the hazard of death.78

Ten Percent Decline in Percent Predicted Forced Expiratory Volume in One Second (FEV1)

Two studies performed univariate analysis to evaluate the relationship between a 10 percent decline in percent predicted FEV1 and mortality, but only one evaluated the multivariate relationship.51,77 The first study found that there was a significant increase in the risk of death for those who had a decrease in percent predicted FEV1 below 10 percent of the predicted value in univariate (RR 1.8, 95 percent CI 1.7 to 2.0) and multivariate (RR 2.0, 95 percent CI 1.9 to 2.2) analysis.51 The second study found that there was a significant increase in the hazard of death for those with a 10 percent decrease in percent predicted FEV1 in univariate (HR 2.1, 95 percent CI 1.5 to 3.0) but not multivariate analysis.77

Forced Expiratory Volume in One Second (FEV1) Z-Score

In one study, the positive predictive value and sensitivity of having an FEV1 Z-score below negative 2 versus a more normal value on the outcome of death or need for transplantation was evaluated.80 The authors suggested a clinically relevant positive predictive value and sensitivity would be 70 percent and 90 percent.80 The positive predictive values for children aged 8, 9, 10, 11 and 12 ranged from 10 to 47 percent and the sensitivities ranged from 33 percent to 76 percent suggesting that having an FEV1 Z-score at or below negative 2 is not a strong predictor of mortality or need for transplantation.80 No differences were seen between those who subsequently died or had a need for transplantation versus those who survived on FEV1 Z-score decline over the previous 2 years in children aged 10 to 12.80

Forced Expiratory Volume in One Second/Forced Vital Capacity (FEV1/FVC) at Baseline

Three studies evaluated the relationship between FEV1/FVC and mortality in univariate but not multivariate analyses.53,65,112 In the first study, there was no significant difference in FEV1/FVC at baseline between those who lived versus those who subsequently died, but the effect size and measures of variance were not reported.112 In the second study, a decline in FEV1/FVC at baseline was not associated with the risk of death (RR 1.00, 95 percent CI 0.98 to 1.03).53 In contrast, in the third trial the FEV1/FVC ratio at baseline was significantly lower in those who survived versus those who subsequently died (MD 0.04, p=0.011).65

Anthropometrics

Twenty-seven studies evaluated the relationship between mortality and various anthropometric measurements and mortality.8,48,50–53,55,57,59–61,63,65–70,72–75,77–80,112 (Appendix Table F2)

Height at Baseline

Five studies evaluated the relationship between height at baseline and mortality using univariate analysis,57,65,69,72,77 while two of these studies used multivariate analysis.57,69 In the first study, those with greater height lived had a reduced risk of death than those with a lesser height based upon univariate analysis, but the effect size was not reported (ES NR, p<0.001) and multivariate analysis (RR 0.033, p<0.0001).57 Like the first study, height was higher in those who lived versus those who subsequently died (MD −3 cm, p=0.073).65 In the third study no significant difference in height at baseline occurred between those who subsequently died and those that lived (MD −0.6in, 95 percent CI −3.44 to 2.24).72 Similar to the third trial and in contrast with the first two trials, the fourth study found that there was no significant difference in height at baseline between those who subsequently died and those who lived (MD −1cm, p=0.30).77

The final study used univariate and multivariate analysis to evaluate the relationship between mortality and the mean height in 1996, when the study began retrospectively reviewing data from the United States Cystic Fibrosis Foundation National Patient Registry in 1996.69 There was a significant association between mean height at baseline and mortality upon univariate analysis but investigators did not report an effect size (ES NR, SS). Multivariate analysis revealed a significant increase in the risk of dying among patients with a higher mean height at baseline than those with a lower mean height at baseline (OR 1.04, 95 percent CI 1.03 to 1.05).69

Height-for-age at baseline

One study evaluated the relationship between mortality and calculated height-for-age at baseline using univariate but not multivariate analysis.73 Those with a higher calculated height at baseline were not significantly more likely to die than those who lived (MD −1, p=0.8).73

Height Quartile

One study evaluated the relationship between mortality and height within the shortest height quartile.77 The hazard of death was not significantly increased for those with a height in the shortest height quartile compared to those with a height above the shortest quartile in univariate (HR 1.4, 95 percent CI 0.9 to 2.4) or multivariate analysis.77

Height Percentile

Two studies evaluated the relationship between height percentile and mortality.8,50 The first study evaluated the baseline height percentile in patients seen in CF clinics in Boston, MA, and Toronto, Canada. Based on univariate analysis, there was not a significant difference in height percentile among those who lived versus those who died at the Boston clinic (MD −1 percent, 95 percent CI −12.29 to 10.29), however those who died at the Toronto clinic had a significantly lower height percentile (MD −10 percent, p<0.05).50 The second study found a significant increase in the hazard of death for males and females at age 5 (males: HR 2.9, 95 percent CI 1.23 to 6.91 and females: HR 4.3, 95 percent CI 2.54 to 7.31) and age 7 (males: HR 6.3, 95 percent CI 2.10 to 18.87 and females: HR 5.8, 95 percent CI 2.53 to 13.11) occurred if the height-for-age was below the 5th percentile.8

Height Z-score

Two studies evaluated the relationship between mortality and height Z-score using univariate analysis,60,70 but only one of the two studies performed multivariate analysis. One study performed univariate, but not multivariate analysis to evaluate relationship between height-for-age Z-score above and below −1.29. The study found that those who had a height-for-age Z-score less then −1.29 did not have a significant increase in the risk of death (RR 4.06, p=0.06).70

The other study evaluated the hazard of death for individuals based on quartile of height Z-score with the lowest quartile further divided above and below the 10th percentile.60 In this study, the hazard of death in the 2 years following Z-score measurement for those with a height Z-score ranging from −0.46 to −1.32 was not significantly increased compared to those with a height Z-score > −0.46 based on univariate (HR 1.4, 95 percent CI 0.9 to 2.1) or multivariate (HR 1.1, 95 percent CI 0.6 to 1.9) analysis.60 The hazard of death in the 2 years following Z-score measurement for those with a height Z-score ranging from −1.33 to −2.21 was significantly increased compared to those with a height Z-score > −0.46 on univariate (HR 1.6, 95 percent CI 1.1 to 2.5) but not multivariate (HR 1.0, 95 percent CI 0.5 to 1.9) analysis.60 The hazard of death in the 2 years following Z-score measurement for those with a height Z-score ranging from −2.22 to −3.25 was significantly increased compared to those with a height z score > −0.46 based on univariate (HR 4.6, 95 percent CI 3.1 to 6.7) but not multivariate (HR 1.9, 95 percent CI 0.9 to 4.1) analysis.60 Finally, the hazard of death in the 2 years following Z-score measurement for those with a height Z-score ≤ −3.26 was significantly increased compared to those with a height Z-score > −0.46 based on univariate (HR 8.8, 95 percent CI 5.9 to 13.1) and multivariate (HR=2.9, 95 percent CI 1.2 to 7.0) analysis.60

Weight

Two studies evaluated the relationship between weight and mortality at the time of evaluation for lung transplantation using univariate, but not multivariate analyses.65,72 The first study found that weight at the time of evaluation for transplant was not significantly different in those who lived compared to those who subsequently died (MD −2.4kg, p=0.200).65 The second study found that weight at the time of evaluation for transplant was not significantly different in those who lived compared to those who subsequently died (MD 6.5 lbs, 95 percent CI −26.61 to 13.61).72

Birth Weight

One study evaluated the relationship between birth weight and mortality using univariate and multivariate analysis.70 This study found that there was a significant reduction in the risk of death for those individuals with a birth weight greater than or equal to 3000 grams when compared to those with a birth weight less than 3000 grams upon univariate (RR 4.06, p=0.01) and multivariate (RR 7, p<0.001) analysis.70

Relative Underweight

One study evaluated the univariate, but not the multivariate relationship between being relatively underweight and mortality.48 The study found that those who lived were less likely to be relatively underweight for age than those who subsequently died, but effect size was not reported (ES NR, p<0.05).48

Weight Percentile

Three studies evaluated the relationship between mortality and weight percentile50,68,112 but only one performed multivariate analysis.112 The first study that evaluated individuals in two different clinics found that there was a significant difference in weight percentile between those who lived and those that subsequently died at both sites (Site 1 MD −25 percent, p<0.001, Site 2: MD −25 percent, p<0.001).50 The second study evaluated the relationship between mortality and weight percentile above and below the fiftieth percentile.68 The hazard of death was significantly increased for those individuals with a weight less than or equal to the fifth percentile when compared to those with a weight greater than fiftieth percentile using univariate analysis (HR 3.9, 95 percent CI 2.1 to 7.3).68 The hazard of death was significantly increased for those individuals with a weight from the fifth to the fifteenth percentile compared to those with a weight greater than the fiftieth percentile based on univariate analysis (HR 2.4, 95 percent CI 1.2 to 4.8).68 Finally, the hazard of death was not significantly increased when those individuals with a weight from the fifteenth to the fiftieth percentile were compared to those with a weight greater than the fiftieth percentile using univariate analysis (HR 1.5, 95 percent CI 0.8 to 2.9).68 In the final study, the weight percentile was significantly lower in patients who died compared to those who lived based upon univariate analysis (MD −10.8 percent, p=0.0001) but lower weight percentile was not a multivariate predictor of mortality.112 Patients who died were more likely to have a weight percentile less than the five at age 18 years than those who lived (MD −39 percent, p=0.0004) based upon univariate analysis, however weight percentile was a multivariate predictor of mortality but study did not report the effect size (ES NR, p<0.0001).112

Percent Predicted Weight

Three studies conducted univariate analysis to evaluate the relationship between percent predicted weight and mortality,50,57,63 but only one of them performed multivariate analysis.50 The first study found a significant univaraite relationship between percent predicted weight and mortality but did not report an effect size.57 The second study did not find a significant difference in percent predicted weight between those who lived and those who subsequently died (MD −3.3 percent, 95 percent CI −6.25 to 12.85).63 The final study found a significant decrease in the hazard of death among those individuals with a higher percent predicted weight compared to those with a lower percent predicted weight in univariate (HR 0.95, 95 percent CI 0.93 to 0.96) but not multivariate analysis (HR 0.99, 95 percent CI 0.98 to 1.00).50

Weight-for-Height

Two studies evaluated the relationship between weight-for-height and mortality using univariate but not multivariate analysis.53,73 The risk of death was significantly decreased (RR 0.96, 95 percent CI 0.92 to 0.99) in those with a higher weight for height compared to those with a lower weight for height.53 In the second study, the weight-for-height at baseline was significant greater for those who lived compared to those who subsequently died (MD −13, p=0.01).73

Percent Predicted Weight-for-Height

One study evaluated the relationship between percent predicted weight-for-height and mortality.51 Both univariate (RR 1.4, 95 percent CI 1.3 to 1.5) and multivariate (RR 1.4, 95 percent CI 1.3 to 1.5) analyses revealed a significant increase in the risk of death for those individuals with a lower percent predicted weight-for-height compared to those individuals with a higher percent predicted weight for height.51

Weight Z-score

Two studies evaluated the relationship between weight Z-score and mortality using univariate and multivariate analyses.60,66 The first study evaluated the relationship between mortality and weight Z-score above and below −0.46 using univariate, but not multivariate analysis.60 In this study, the hazard of death was not significantly increased for those with a weight Z-score between −0.49 and −1.25 when compared to those with a weight Z-score greater than −0.49 (HR 1.2, 95 percent CI −0.7 to 2.1).60 The hazard of death was significantly increased in those with a weight Z-score between −1.26 and −1.98 when compared to those with a weight Z-score greater than −0.49 (HR 2.8, 95 percent CI 1.7 to 4.4).60 The hazard of death was significantly increased in those with a weight Z-score between −1.98 and −2.74 when compared to those with a weight Z-score greater than −0.49 (HR=7.8, 95 percent CI 5.0 to 12.2).60 Finally, the hazard of death was significantly increased in those with a Z-score ≤ −2.75 when compared to those with a Z-score greater than −0.49. (HR 16.4, 95 percent CI 10.5 to 25.6).60

The second study found a statistically significant univariate relationship between mortality and weight for age Z-score but did not report an effect size (ES NR, SS), while multivaratie analysis revealed a non-significant decrease in the odds of death for those with a higher weight for age Z-score compared to those with a lower weight for age Z-score (OR 0.75, NS).66

Body Mass Index (BMI) ≤ 16 versus ≥18.6

One study evaluated the relationship between mortality and a BMI less than or equal to 16 or greater than or equal to 18.6 using univariate, but not multivariate analysis.52 In this study there was no significant difference in the risk of death for those with a BMI less than or equal to 16 when compared with those who had a BMI greater than or equal to 18.6 (RR 1.6, 95 percent CI 0.8 to 3.1).52

Body Mass Index (BMI) at Baseline

Three studies evaluated the relationship between BMI at baseline and mortality using univariate and multivariate analyses.59,61,79 The first study found that BMI at baseline was significantly greater among those who lived compared to those who died using univariate analysis (MD −1.9 kg/m2, p=0.001) but BMI at baseline was not a significant multivariate predictor of mortality (ES NR, NS).59 The second study did not report an effect size for univariate or multivariate analysis, but did report a significant univariate (ES NR, p=0.05) and multivariate, but the effect size was not reported (ES NR, SS) relationship between BMI and mortality.61 In the last study individuals who lived had a significantly higher BMI at baseline than those who died (MD −1.5kg/m2, p=0.008), but baseline BMI was not a multivariate predictor of mortality, but the effect size was not reported (ES NR, p=0.31).79

Body Mass Index (BMI) Prior to Intensive Care Unit Admission

Three studies evaluated the relationship between mortality and BMI prior to admission to the intensive care unit for pulmonary exacerbation using univariate but not multivariate analyses.74,75,78 The first study found that there was a significant increase in the risk of death among those with a BMI less than 18 at the time of admission to the intensive care unit compared to those with a BMI greater than 18 at the time of admission (RR 3.25, 95 percent CI 1.27 to 3.25).74 The second study found a non-significant decrease in the hazard of death for those individuals with a lower BMI at the time of admission to the Intensive Care Unit compared to those with a higher BMI at the time of admission (HR 0.87, 95 percent CI 0.69 to 1.11).75 A final study found that there was no significant increase in the hazard of death for those patients admitted to the intensive care unit that had a lower BMI when compared to those with a higher BMI (HR 0.95, 95 percent CI 0.80 to 1.13).78

Body Mass Index (BMI) at Time of Transplant Evaluation

Two studies evaluated the relationship between mortality and BMI using univariate but not multivariate analysis.72,77 The first study found no significant difference in BMI at the time of evaluation for lung transplant when individuals who lived were compared to those who subsequently died using univariate analysis (MD −1.26 kg/m2, 95 percent CI −3.91 to 1.39).72 The second study found no significant increase in hazard of death for those with a lower BMI than those with a higher BMI at the time of evaluation for lung transplant (HR 1.0, 95 percent CI 0.9 to 1.1).77

Percent Ideal Body Weight (IBW)

One study evaluated the relationship between percent ideal body weight in patients at the time of listing for lung transplant and mortality using univariate but not multivariate analysis, and found the percent IBW was not significantly higher in those who lived versus those who subsequently died (MD 1 percent, p=0.685).65

Percent Ideal Body Weight (IBW) at Baseline

One study evaluated the relationship between percent ideal body weight at baseline and mortality using univariate and multivariate analysis.67 The study found that there was a significant decrease in the risk of death for those with a higher percent IBW at baseline compared to those with a lower percent IBW at baseline upon univariate (RR 0.955, 95 percent CI 0.944 to 0.967) and multivariate (RR 0.968, 95 percent CI 0.947 to 0.99) analysis.67

Percent Ideal Body Weight (IBW) ≤85 percent

One study evaluated the relationship between percent ideal body weight less than 85 percent and mortality.67 In the study there was a significant increase in the hazard of death for those individuals with a percent IBW less than 85 percent when compared to those with a percent IBW greater than 85 percent (HR 2.64, 95 percent CI 1.85 to 3.75).67

Percent Ideal Body Weight (IBW) Evaluated by Percent

One study evaluated the relationship between percent ideal body weight and mortality using univariate, but not multivariate analysis.60 The study found that the hazard of death was not significantly increased for individuals with a percent IBW between 98 and 104.9 when compared to those with a percent IBW greater than 105 (HR 0.9, 95 percent CI 0.6 to 1.5).60 The hazard of death was significantly increased for individuals with a percent IBW between 90 and 97.9 when compared to those with a percent IBW greater than 105 (HR 1.6, 95 percent CI 1.1 to 2.3).60 The hazard of death was significantly increased for individuals with a percent IBW between 84 and 89.9 when compared to those with a percent IBW greater than 105 (HR 3.2, 95 percent CI 2.2 to 4.7).60 The hazard of death was significantly increased for individuals with a percent IBW less than 84 when compared to those with a percent IBW greater than 105 (HR 7.1, 95 percent CI 5.0 to 10.2).60

Protein Turnover

No studies reported the relationship between protein turnover and mortality in CF patients.

Exercise Tolerance

A total of 10 studies evaluated the link between various measures of exercise tolerance and mortality.52–54,58,59,63,65,72,76,77 (Appendix Table F3)

Walk Testing

A total of nine studies evaluated the link between exercise tolerance and mortality.53,54,58,59,63,65,72,76,77

Five studies evaluated the link between 6 minute walk testing and mortality.54,58,63,65,77 Three studies evaluated the relationship between distance walked in meters and mortality using univariate but not multivariate analyses.54,58,63 In these studies, patients who subsequently died qualitatively walked a lesser distance (ranging from 43 to 137 meters less in distance walked) than those who survived, but walking distance was only a univariate predictor of mortality in one of the three studies (MD −137.4, p=0.016). In the fourth study, the univariate and multivariate relationship between distance walked, in 50 meter increments, and mortality were evaluated.65 For each 50 meter increase in the six minute walk distance, the risk of death decreased by 27 percent (RR 0.73, 95 percent CI 0.62 to 0.87) in univariate and 31 percent (RR 0.69, 95 percent CI 0.57 to 0.84) in multivariate analysis. In the same study, for every five percent incremental increase in distance walked, the risk of death decreased by 18 percent (RR 0.82, 95 percent CI 0.72 to 0.94) but multivariate analysis was not conducted.65 In the fifth study, individuals with a lesser six minute walk distance did not have an increased hazard of death than those with a greater six minute walk distance (HR 1.0, 95 percent CI 0.99 to 1.0) upon univariate analysis and no multivariate analysis was conducted.77

In a similar study, the univariate relationship between exercise tolerance after a 12 minute walk and mortality was explored but multivariate analysis was not conducted.53 When comparing patients who walked a distance above the median of 540 meters against patients who walked a distance below the median, the relative risk of death was not significantly increased (RR 0.89, 95 percent CI 0.41 to 1.95).53

Percent Predicted Peak Oxygen Uptake (VO2-peak)

Two studies evaluated the relationship between percent predicted peak oxygen uptake during exercise testing.52,59

In the first study, those who lived had a significantly greater percent predicted peak oxygen uptake than those who subsequently died (MD −12.9 percent, p=0.022) but no significant multivariate relationship was seen.59 Unfortunately, the effect size and measure of variance were not reported for the multivariate analysis.59

In the second study, both univariate and multivariate analysis were performed to evaluate the relationship between having a peak oxygen uptake less than or equal to 58 percent versus greater than or equal 82 percent during exercise testing and mortality.52 The relative risk of death was higher subjects with peak oxygen uptake of less than or equal to 58 percent versus those with values greater than or equal to 82 percent during exercise testing in univariate (RR 6.4, 95 percent CI 2.6 to 15.7) and multivariate (RR 3.2, 95 percent CI 1.2 to 8.6) analysis, respectively.52

Peak Oxygen Uptake (VO2-peak)

Two studies evaluated the relationship between maximum oxygen uptake (VO2-peak) during exercise testing and mortality using univariate but not multivariate analysis.72,76 In the first study, there was no significant difference in maximum oxygen uptake during exercise testing between those who subsequently died and those who survived (MD −0.171 L/min, 95 percent CI −1.85 to 2.19).72 In the second study, no relationship was seen between increasing peak oxygen uptake and mortality (HR 0.953, 95 percent CI 0.865 to 1.051) in initial testing but a univariate relationship was seen during final testing where for every 1mL/min/kg increase in peak oxygen uptake, the hazard of death was (MD 0.845, 95 percent CI 0.757 to 0.944).76

Percent Predicted Peak Work Rate (Wpeak)

One study evaluated the relationship between the percent predicted peak work rate during exercise testing and mortality using both univariate and multivariate analysis.59 Those patients who lived had a significantly greater percent predicted peak work rate during exercise testing than those who subsequently died (MD −18.1 percent, p=0.015) in univariate analysis but peak work rate was not a multivariate predictor of mortality.59

Minute Ventilation/Peak Oxygen Uptake (VE/VO2)

One study evaluated the relationship between the ratio of minute ventilation to peak oxygen uptake during exercise testing and mortality using both univariate and multivariate analysis.59 The study found that those who died had a statistically greater VE/VO2 during exercise testing than those who lived (MD 6.3, p=0.002), and that VE/VO2 was not a multivariate predictor of mortality.59

Peak Minute Ventilation (VEpeak)

One study evaluated the relationship between the peak minute ventilation during exercise testing and mortality using both univariate and multivariate analysis.59 The peak minute ventilation was greater in those who lived than those who subsequently died (MD −8.1 L/min, p=0.04) but peak minute ventilation was not a significant multivariate predictor of mortality.59

Bone Mineralization

No studies reported the relationship between bone mineralization and mortality in CF patients.

Health-Related Quality of Life
Pulmonary Function

A total of 14 studies evaluated the link between pulmonary function and various scales of HRQoL.82–93,95–98 (Appendix Table F4)

Alltagsleben (Every Day Life)

One study evaluated the link between pulmonary function and the Alltagsleben questionnaire, a scale developed for German-speaking patients.86 Staab and colleagues evaluated 89 adolescents and adults with CF.86 Staab and colleagues analyzed results in two different hierarchical regression analysis models, the first of which included subjective health perception variables and the other which included ways of coping variables.86 In univariate analysis, there was a significant positive relationship between FEV1 and Alltagsleben scores (r=0.31, p<0.01 for model 1 where n=83, and r=0.36, p<0.001 in model 2 where n=84).86 From the publication, it is not specified whether the values for FEV1 are absolute or percent predicted, as units of measure were not reported; however, in looking at the mean values of FEV1 and their range in the patients studied, it appears to be percent predicted FEV1.86

Upon multivariate analysis, FEV1 was no longer statistically significant in model 1 (β=0.12, p-value not reported), but retained statistical significance in model 2 (β=0.24, p<0.05).86

Child Health Questionnaire

The univariate relationship between pulmonary function and CHQ was evaluated in two studies but multivariate analyses were not conducted.89,92 Powers and colleagues evaluated 24 adolescents with CF during a routine CF clinic visit.89 In these patients, there was a significant positive univariate relationship between percent predicted FEV1 and the domains of general health, role/social-physical, and bodily pain (r=0.73, 0.47, and 0.42 respectively, p<0.05 for all).89 Nonsignificant positive correlations were found between percent predicted FEV1 and domains of physical functioning, role/social-emotional, mental health, family activities, and self-esteem (range of r values were from 0.24 to 0.39, p-values not reported).89 Nonsignificant negative univariate correlations were found between percent predicted FEV1 and domains of role/social-behavior and behavior problems (range of r values were from −0.21 to −0.04, p-values not reported).89

In another evaluation of 36 patients, ranging in age from 10 to 15.5 years, there was no univariate relationship found between percent predicted FEV1 to any of the 12 subscores of the CHQ (p-values not reported), with the exception of family cohesion subscore (r=0.37, p=0.05).92

Cystic Fibrosis Quality of Life Questionnaire

One study evaluated the link between pulmonary function and the CFQoL scale.90,91 Adults and adolescents with CF were surveyed.90,91 Univariate results are presented separately for males and females.90 In females, there was a significant positive correlation between percent predicted FEV1 and seven domains of the CFQoL (emotional functioning, relationships, physical functioning, body image, chest symptoms, career issues, and treatment issues; r values ranged from 0.17 to 0.60, p<0.05 for all), while there was no significant relationship between FEV1 and the two domains of concerns for the future or social functioning (p-values not reported).90 In males, there was a significant relationship between percent predicted FEV1 in all domains (r values ranged from 0.22 to 0.50, p<0.05 for all) except the domain for relationships (p-value not reported).90

Multivariate analysis combined data for participants regardless of gender.91 Upon multivariate analysis, percent predicted FEV1 was significantly associated with seven domains (physical functioning, social functioning, treatment issues, chest symptoms, emotional functioning, concerns for the future, and interpersonal relationships; β values ranged from 0.12 to 0.29, (p-values not reported) and was not significantly associated with body image or career concerns (p-values not reported).91

Cystic Fibrosis Questionnaire

Three studies evaluated the univariate link between pulmonary function and CFQ but multivariate analyses were not conducted.93,96–98 All three studies found a significant positive univariate relationship between percent predicted FEV1 and physical domain (r values ranged from 0.27 to 0.57, p<0.05 for all) and health perceptions domain (r values ranged from 0.38 to 0.51, p<0.05 for all).93,96–98 Havermans and colleagues did not report results for any of the remaining domains of the CFQ.97,98 Both Quittner and colleagues and Riekert and colleagues found a significant positive univariate relationship between percent predicted FEV1 and seven other CFQ domains (role, vitality, social, body image, eating, respiratory, and weight; r values ranged from 0.23 to 0.41, p<0.01 for all).93,96 Quittner and colleagues additionally found a significant positive univariate correlation between percent predicted FEV1 and the emotion domain (r=0.28, p<0.01), while Riekert and colleagues did not (r=0.20, p-value not reported).93 Percent predicted FEV1 was also positively associated with the treatment domain in the study by Riekert and colleagues (r=0.32, p<0.01), but was not significantly correlated in the study by Quittner and colleagues (r=0.11, p-value not reported).93,96 The digestive domain was not associated with percent predicted FEV1 in either study (r values ranged from 0.01 to 0.03, p-values not reported).93,96

EuroQoL-5D

The link between pulmonary function and EQ-5D was reported in one study.87 At baseline evaluation of 39 adults with CF, percent predicted FEV1 was significantly positively associated with EQ-5D on univariate analysis (Spearman’s ρ=0.427, p=0.017).87

After 1 year, the EQ-5D was readministered to patients and there was a significant multivariate relationship between percent predicted FEV1 at baseline and EQ-5D index after 1 year (β=0.000, p=0.005).87

Medical Outcomes Short Form-36

Two studies evaluated the univariate link between pulmonary function and SF-36 but multivariate analyses were not conducted.87,88 In adults with CF (n=39), there was a significant positive univariate relationship between percent predicted FEV1 and the physical composite score (Spearman’s ρ=0.396, p=0.025).87 Abbott and colleagues evaluated English CF patients aged 14–18 years (n=58) and German patients aged 13 to 17 years (n=26).88 In both English and German patients, there was a significant positive univariate relationship between percent predicted FEV1 and the physical functioning subscore (r=0.39, p<0.003 and r=0.43, p<0.03, respectively).88 Neither population showed significant relationships between percent predicted FEV1 and the remaining domains of the SF-36 (physical role limitation, social functioning, mental health, mental role limitation, energy and vitality, general health perceptions, changes in health; p-values not reported).88

Nottingham Health Profile

One study reported on the univariate link between pulmonary function and NHP but multivariate analysis was not conducted.84 In clinically stable adults with CF (age over 16 years), there was a statistically significant negative univariate relationship between percent predicted FEV1 and all subscores of NHP (r values ranged from −0.51 to −0.15, p<0.05 for all), with higher percent predicted FEV1 indicating better HRQoL.84

Quality of Well-Being Scale

Two studies evaluated the univariate relationship between pulmonary function and QWB scores but multivariate analyses were not conducted.82,83 Orenstein and colleagues evaluated CF patients aged 7 to 36 years (n=44) and found a statistically significant positive relationship between absolute FEV1 and QWB (r=0.55, p<0.0001) on univariate analysis.82

Czyzewski and colleagues found no relationship between pulmonary function and QWB in their study of patients under age 18 (r= −0.07 and 0.001 for percent predicted FEV1 and percent predicted FVC, respectively, p-values not reported).83

Questions on Life Satisfaction

The relationship between pulmonary function and the Questions on Life Satisfaction Scale was evaluated in one study.95 Results of univariate analysis were not reported. Upon multivariate analysis of patients aged at least 15 years with CF, neither percent predicted FEV1 at second clinic visit nor the change in percent predicted FEV1 between two clinic visits were significantly associated with HRQoL scores (p-values not reported).95

Sickness Impact Profile

One study evaluated the univariate link between pulmonary function and SIP but multivariate analysis was not conducted.85 Upon univariate analysis, percent predicted FEV1 negatively correlated with overall SIP and physical subscores but was not statistically significant (r= −0.33 and −0.40, respectively, p-values not reported), and was nonsignificantly positively correlated with psychosocial subscore (r=0.05, p-value not reported).85

Anthropometrics

A total of 10 studies evaluated the link between anthropometric measures and various scales of HRQoL.84,86–88,90–95,97,98 (Appendix Table F5)

Alltagsleben (Every Day Life)

One study evaluated HRQoL using the “Alltagsleben” questionaire, a scale developed for German-speaking patients.86 Staab and colleagues evaluated 89 adolescents and adults with CF.86 Staab and colleagues analyzed results in two different hierarchical regression analysis models, the first of which included subjective health perception variables and the other which included ways of coping variables.86 Univariate analysis showed no significant relationship between percent IBW and HRQoL (r=0.11 in model 1 where n=83, and r=0.10 in model 2 where n=84).86 In multivariate analysis, no significant association between percent IBW and subjective health perception or ways of coping (β=0.05 and −0.11 in models 1 and 2, respectively, p-values not reported).86

Child Health Questionnaire

One study evaluated the univariate relationship between anthropometrics and the CHQ scale but multivariate analysis was not conducted.92 In 36 patients ranging in age from 10 to 15.5 years, there was no univariate relationship found between either height-for-age or weight-for age Z-scores and any of the 12 subscores of the CHQ (p-values not reported), with the exception of weight-for-age Z-score and the general health perception subscore (r=0.36, p=0.03).92

Cystic Fibrosis Quality of Life Questionnaire

One study evaluated the association between anthropometrics and the CFQoL scale. Adults and adolescents with CF were surveyed.90,91 Univariate results are presented separately for males and females.90 The only domain that had a significant positive relationship in both males and females was the relationship between BMI and the CFQoL body image subscore (r=0.34 and 0.55 respectively, p=0.001 for both).90 Males additionally experienced a significant positive relationship between BMI and chest symptoms (r=0.21, p=0.02), while females did not (p-value not reported).90 Females also showed a significant positive relationship between BMI and concerns for the future (r=0.20, p=0.02), while males did not (p-value not reported).90 All other domains of CFQoL (physical functioning, social functioning, treatment issues, emotional functioning, social functioning, treatment issues, emotional functioning, interpersonal relationships, and career concerns) were not significant for both males and females (p-values not reported).90

For multivariate analysis, all patients were analyzed as a single group, regardless of gender.91 Higher BMI was associated with the body image subscore (β=3.4, 95 percent CI 2.1 to 4.6), but all other domains were not significantly associated (p-values not reported).91

Cystic Fibrosis Questionnaire

Three studies evaluated the link between anthropometrics and CFQ.93,97,98 Two studies reported results on each of the twelve subscores of the CFQ in adults and adolescents with CF.93,97,98 Both studies found a significant positive univariate relationship between BMI and body image domain and eating domain, with higher BMI correlating with higher body image score (r values ranged from 0.28 to 0.38, p<0.05 for all) and higher eating domain scores (r values ranged from 0.16 to 0.44, p<0.05 for all).93,98 Higher BMI was also associated with higher weight domain scores in both studies (r values ranged from 0.43 to 0.47, p<0.01 for all).93,98 Quittner and colleagues additionally found a significant relationship between BMI and health perceptions domain scores (r=0.14, p<0.05).93 No other domains in Quittner’s analysis were statistically significant,93 and results from the remaining domains in the study by Havermans and colleagues were not reported.98

Multivariate analysis was conducted in one study, which reported CFQ results in three dimensions (physical, emotion, and social).94 Koscik and colleagues found that neither adequate weight gain within 2 years of diagnosis nor BMI Z-score greater than −1 were predictors of social or emotion dimension scores upon multivariate analysis of 45 CF patients aged 8 to 18 years.94 However, adequate weight gain was associated with improvements in the CFQ physical dimension (model p=0.04 after adjusting for age), though BMI Z-score was not (model p=0.52 after adjusting for age).94

EuroQoL-5D

One study evaluated the relationship between anthropometrics and EQ-5D.87 At the baseline evaluation of adults with CF (n=39), there was no significant univariate relationship between BMI and EQ-5D VAS (p-value not reported).87 After 1 year, the EQ-5D was re-administered to patients and there was a significant negative multivariate relationship between BMI at baseline and EQ-5D index after 1 year (β= −0.002, p=0.005).87

Medical Outcomes Short Form-36

Two studies evaluated the univariate association between anthropometrics and SF-36, multivariate analyses were not conducted.87,88 In one study comprised of adult patients with CF (n=39), there was no significant univariate relationship between BMI and either the SF-36 physical composite score or mental composite score (p-value not reported).87

Abbott and colleagues evaluated English CF patients aged 14–18 years (n=58) and German patients aged 13 to 17 years (n=26).88 In both populations, there was no significant univariate relationship between BMI and any of the SF-36 subscores (physical functioning, physical role limitation, social functioning, mental health, mental role limitation, energy and vitality, general health perception, or changes in health) (p-values not reported).88

Nottingham Health Profile

One study evaluated the univariate association between anthropometrics and HRQoL, using the NHP, multivariate analysis was not conducted.84 In 240 patients aged over 16 years, there was a significant univariate negative correlation between BMI and NHP energy subscore, NHP sleep subscore, and physical mobility subscore (p<0.001, p<0.05, p<0.0001, respectively), meaning that higher BMI represents better HRQoL.84 The relationship between the remaining NHP subscores (pain, emotion, and social isolation) were not significant (p-values not reported).84

Questions on Life Satisfaction

One study reported on the relationship between anthropometrics and the Questions on Life Satisfaction scale.95 In 108 adult and adolescent patients with CF, there was no relationship between BMI and HRQoL in multivariate analysis (p>0.15).95

Protein Turnover

No studies reported the relationship between protein turnover and HRQoL in CF patients.

Exercise Tolerance

Two studies evaluated the relationship between exercise tolerance and HRQoL.82,85 In pediatric and adult patients evaluated with bicycle ergometer testing and the Quality of Well-Being Scale (QWB), there was a statistically significant relationship between Vo2-peak and QWB scores, with higher exercise capacity relating to better QWB scores (p<0.01).82,85 (Appendix Table F6)

In adult patients evaluated with bicycle ergometer testing and the Sickness Impact Profile (SIP), maximal workload (Wpeak) negatively correlated with SIP Overall Score and Physical Subscore (p<0.05 and p<0.01, respectively), indicating that greater exercise capacity related to better HRQoL.85 There was no statistically significant relationship between Wpeak and SIP Psychosocial Subscore.85 (Appendix Table F6)

Bone Mineralization

No studies reported the relationship between bone mineralization and HRQoL in CF patients.

Bone Consequences

Pulmonary Function. One study evaluated the association between pulmonary function and important bone consequences.99 There was no significant relationship between either FEV1 or FVC and bone fracture, with patients who had experienced bone fracture showing no significant difference in pulmonary function than those who had not experienced fracture.99 (Appendix Table F7)

Anthropometrics. The study by Aris and colleagues was also the only ones to evaluate the association between anthropometrics and bone fracture.99 There was no significant difference in BMI between patients who has experienced bone fracture and those who did not.99 (Appendix Table F8)

Protein Turnover. No studies reported the relationship between protein turnover and bone consequences in CF patients.

Exercise Tolerance. No studies reported the relationship between exercise tolerance and bone consequences in CF patients

Bone Mineralization. No studies reported the relationship between bone mineralization and bone consequences in CF patients.

Discussion

With the limited amount of evidence regarding the impact of rhGH on final health outcomes, it is important to determine if the outcomes affected by rhGH would ultimately predict final health outcomes like HRQoL, bone consequences, and mortality. Therefore, studies which follow the natural progression of CF were sought to determine these linkages. While univariate analysis provides some insight into the relationship between an intermediate and final health outcome, the associations can be greatly impacted by confounders and thus provides weaker evidence. Multivariate analysis allows the determination of a variable’s predictive ability, independent of other possible confounding variables. As such, multivariate predictors of an outcome provide more compelling evidence of an association.

The relationship between absolute change in FVC and mortality is weak with one study showing an association but the majority of studies finding no association. The relationship between percent predicted FVC and mortality is also weak, with three studies showing an association, but the majority of studies showing no association. In contrast, half of the studies reported an association between decline in FVC and mortality. In KQ1, we found that rhGH significantly increased the absolute measures of FVC and percent predicted FVC. However, because there was not a strong link between measures of FVC and mortality in KQ3, we cannot be confident in our ability to extrapolate improvements in FVC associated with rhGH therapy to improvements in survival.

The relationship between absolute changes in FEV1 and mortality is controversial with some limited studies showing an association and others not finding an association. The relationship between percent predicted FEV1 and mortality is much stronger with many more trials evaluating this association and a majority finding an association between higher percent predicted FEV1 and improved survival. A model describing the relationship between pulmonary function and survival was consistent with our findings, showing that longer survival was associated with higher FEV1 and lower rates of FEV1 decline.116 In KQ 1, we found that rhGH significantly increased absolute measures of FEV1, but did not significantly increase percent predicted FEV1. As such, we are not as confident in our ability to extrapolate improvements in FEV1 associated with rhGH therapy to improvements in survival.

Only one of three studies found an association between the FEV1/FVC ratio and mortality but none of the trials included in KQ1 evaluated the effects of rhGH on FEV1/FVC.

Height at baseline was not strongly associated with mortality with only two studies reporting an association and the majority of studies reporting no association. One study evaluated the relationship between height for age at baseline and mortality and found no association with mortality. Only one study evaluated the relationship between height quartile and mortality and found no association with mortality. A single study evaluated the relationship between height quartile and mortality finding no association. There was a univariate association between mortality in two of the three studies that evaluated mortality and height percentile but there was no multivariate association. Three studies evaluated the relationship between height Z-score and mortality and found an association for those individuals with lower Z-scores compared to those with higher Z-scores in the majority of univariate and multivariate analyses. In KQ 1, we found that rhGH significantly increased measures of height from baseline, and change from baseline in Z-score. However, because there was not a strong link between measures of height and mortality in KQ3, we can not be confident in our ability to extrapolate improvements in height associated with rhGH therapy to improvements in survival.

The relationship between being relatively underweight and mortality was only evaluated by one study with a univariate association reported. The relationship between weight percentile and mortality was evaluated by three studies and all three studies found a univariate association with mortality, while only one found a multivariate relationship. The majority of studies evaluating the relationship between mortality and percent predicted weight found a univariate relationship; however, none reported a multivariate relationship. Weight Z-score was a significantly associated with mortality in univariate analyses performed by two studies, but no multivariate associations were found. In the only study that evaluated percent predicted weight-for-height, there was an association upon univariate and multivariate analysis. Weight-for-height was associated with mortality in univariate analysis in two studies, but neither found a multivariate association. Finally, there was only an association between mortality and weight in one of the three studies evaluating weight as predictor of mortality. In KQ1, we found that rhGH significantly increased absolute measures of weight from baseline, weight percentile and trended toward improvements in Z-score. However, because there was not a strong link between measures of weight and mortality in KQ3, we cannot be confident in our ability to extrapolate improvements in weight associated with rhGH therapy to improvements in survival.

One study comparing BMI below 16.0 and above 18.6 found no univariate or multivariate association with mortality. All of the studies evaluating change in BMI from baseline found a significant univariate association with mortality, but only one found a multivariate association. In studies evaluating BMI prior to intensive care unit admission, only one found a univariate association with mortality. Finally, none of the studies evaluating BMI at the time of evaluation for transplant found an association with mortality. In KQ 1, we found that rhGH significantly increased the absolute measures of BMI from baseline. As such, we are not as confident in our ability to extrapolate improvements in BMI associated with rhGH therapy to improvements in survival.

The only study that evaluated the relationship between percent IBW found no association with mortality. In contrast, the only study that evaluated the relationship between percent IBW at baseline and mortality found a univariate and multivariate association with mortality. Percent IBW below 85 percent was associated with mortality on univariate, but not multivariate analysis. Similarly, when percent IBW ranging from 84 percent to 97.9 percent compared to <84 percent there was a significant univariate but not multivariate association with mortality. In KQ 1, we found that rhGH significantly improved percent of ideal body weight but we are not as confident in our ability to extrapolate improvements in ideal body weight associated with rhGH therapy to improvements in survival.

Six-minute walk distance was associated with mortality in only one of the studies evaluating this outcome and none of the studies found a multivariate association. Additionally, there was no association between 12-minute walk distance and mortality when univariate analysis was conducted. In one study that evaluated exercise tolerance in 50 meter increments there was an association with mortality in univariate and multivariate analysis. There was a univariate association between 5 percent incremental increases in walk distance and mortality. However, the impact of rhGH therapy on exercise tolerance in CF patients has not been firmly established.41

Two studies evaluating percent predicted peak oxygen uptake during exercise testing found a univariate association with mortality, but only one reported a multivariate association. In both studies, evaluating the relationship between mortality and peak oxygen during exercise testing uptake there was no association between mortality and initial peak oxygen uptake, but there was a univariate association between final peak oxygen uptake and mortality. The study that evaluated the relationship between percent predicted work rate and during exercise testing and mortality found a univariate association with mortality, but was not associated on multivariate analysis. In KQ1, we found there was no significant improvement in work rate. Even if rhGH impacted work rate, we would not be confident in our ability to extrapolate improvements to improvements in survival.

The study that evaluated the association between the ratio of VE to VO2 and mortality found a univariate but not multivariate association. One study found that peak minute ventilation was associated with mortality in univariate but not multivariate analysis. None of the trials included in KQ1 evaluated the effects of rhGH on the ratio of VE to VO2.

There are several limitations to the applicability of studies evaluating the link between intermediate outcomes and HRQoL to patients receiving rhGH therapy in CF. Most studies were cross-sectional and may not elucidate the relationship between clinical status at the time of rhGH administration and HRQoL throughout the patient’s life. One study113 did specifically evaluate this but the results are only available in abstract form even though the abstract was published in 2006. These studies evaluate the innate effect of FEV1 or anthropometrics on HRQoL and do not account for the potential treatment burden of using rhGH that could attenuate HRQoL improvements. In many studies, relationships were made between intermediate outcomes and individual domains of HRQoL scales, not to the overall HRQoL scales. In these studies, an association between improvements in an intermediate outcomes and several HRQoL subscales may potentially be interpreted as an association with overall HRQoL, but this may be an erroneous conclusion. Finally, only eight studies explicitly state that determining the relationship between intermediate outcomes and HRQoL was an primary objective of their study84,85,87,89–91,94–96 with other studies stating primary objectives such as validity assessment either of the tool in a CF population,82,83,93 or comparing HRQoL in CF patients to generally healthy persons.84,88

While numerous studies evaluated the association between higher FEV1 and HRQoL, all of the trials where it was specified evaluated percent predicted FEV1 instead of absolute FEV1. This is problematic since our analysis in KQ 1 suggests that rhGH improved absolute FEV1 but may not impact percent predicted FEV1. As such, it makes it difficult to link the impact that rhGH has on absolute FEV1 with HRQoL. That being said, three of four multivariate analyses found some link between improvements in FEV1 and improvements in HRQoL.86,87,91,95

The multivariate relationship between anthropometrics and HRQoL was assessed in five studies.86,90,91,95,113,117 Most studies showed no relationship between improved measures of weight and improvements in HRQoL. The lack of association may occur because adolescents and adults prefer being underweight.84,87 However, one study showed a positive correlation between adequate weight gain at diagnosis and the physical domain of CFQ94 while another study showed a negative relationship between BMI and EQ-5D.87 This seemingly contradictory information is difficult to interpret and may be attributable to several factors. Koscik and colleagues evaluated the impact of adequate weight gain within the first 2 years of CF diagnosis on HRQoL in later childhood,94 while Johnson and colleagues evaluated adult patients in a cross-sectional survey with a 1 year followup.87 It is possible that adults view their HRQoL differently than children and the impact body weight may have also differs between those populations. The difference might also be due to the specific use of a physical domain of a scale versus an assessment of overall HRQoL in the other study. Finally, an absolute measure of weight, the BMI, might not be as important in determining HRQoL as how a patient perceives their weight growth over a period of time.

Only univariate evidence is available to evaluate the link between exercise tolerance and HRQoL. Both studies found that greater exercise tolerance is significantly related to improvements in HRQoL.82,85 Based on very limited data, the improvements in exercise tolerance may not impact psychosocial functioning. From the analysis in KQ1, rhGH qualitatively improved measures of exercise tolerance, although none of the results reached statistical significance.

Patients with CF are inherently at risk for adverse bone consequences including osteopenia, osteoporosis, and fractures because of vitamin D malabsorption, poor nutritional status, delayed pubertal maturation, and physical inactivity.107 In the one study to evaluate the association between either FEV1, FVC, or BMI and bone fracture, no associations were found. This was a relatively small study and only evaluated bone fracture. Whether associations would have been seen between these factors and osteopenia or osteoporosis is not known.99 In addition, aside from the impact of rhGH on pulmonary function and anthropometrics, rhGH does improve bone mineralization which might directly impact adverse bone consequences.

Key Question 4. In patients with CF, what is the frequency of nonmalignant serious adverse effects resulting from treatment with rhGH? Adverse effects of interest include, but are not limited to, glucose intolerance, diabetes, and hyperglycemia

Key Points

  • rhGH therapy does not impact A1c in CF patients.
  • In CF patients, rhGH therapy significantly increases fasting blood glucose concentrations but does not significantly alter random, postprandial and stimulated blood glucose concentrations.
  • Most CF patients receiving rhGH did not develop glucose intolerance or diabetes over the duration studied (6 to 12 months).
  • In CF patients receiving rhGH, injection site reactions are a rare and insignificant adverse effect.
  • CF patients on rhGH may rarely experience a transient increase in liver transaminases.
  • Study withdrawals were rare in trials evaluating rhGH in CF patients.

Detailed Analysis

Study Design and Population Characteristics

The studies which report nonmalignant adverse effects of rhGH therapy are derived from the same set of studies used to evaluate Key Question 1 and are summarized in Table 3Table 7.

Outcome Evaluations

Glucose Parameters

Five controlled trials, summarized in, Table 2Table 23 reported information on various glucose parameters including fasting, stimulated, and postprandial glucose concentrations as well as glycosylated hemoglobin (A1c) levels in patients with CF.4,24,27,34,35

Table 22. Baseline glucose parameters in controlled trials evaluating rhGH in CF patients.

Table 22

Baseline glucose parameters in controlled trials evaluating rhGH in CF patients.

Table 23. Change from baseline in glucose parameters in controlled trials evaluating rhGH in CF patients.

Table 23

Change from baseline in glucose parameters in controlled trials evaluating rhGH in CF patients.

Two single-arm observational studies, summarized in Table 24, also reported glucose parameters in patients on rhGH therapy while other studies only reported a general statement regarding the impact on glucose.41–43,45,46

Table 24. Change from baseline in glucose parameters in single-arm observational studies evaluating rhGH in CF patients.

Table 24

Change from baseline in glucose parameters in single-arm observational studies evaluating rhGH in CF patients.

Glycosylated Hemoglobin. Hemoglobin A1c levels were specifically reported in two controlled trials and were amenable to quantitative synthesis.24,34 Upon statistical pooling, hemoglobin A1c did not show statistically significant differences between patients with CF receiving rhGH and those without treatment (WMD −0.10 percent, 95 percent CI −0.40 to 0.20 percent). (Figure 23) There were too few trials to conduct evaluations of publication bias and statistical heterogeneity. However, both trials showed a similar direction and magnitude of effect.

Figure 23. KQ4—meta-analysis of change from baseline in A1c in CF patients treated with rhGH.

Figure 23

KQ4—meta-analysis of change from baseline in A1c in CF patients treated with rhGH. Legend: A1c=glycosylated hemoglobin; CF=cystic fibrosis; rhGH=recombinant human growth hormone Note: The squares represent individual point estimates. The size (more...)

A1c was also evaluated in two prospective, single-arm, observational studies which administered rhGH therapy to all participants. These observational studies reported the following qualitative changes from baseline, though the statistical significance is uncertain: B0.2±0.8 percent and 0.1±0.5 percent.42,45

Random Blood Glucose. Random blood glucose concentrations were evaluated in three controlled trials but quantifiable data was only reported in one trial, precluding quantitative synthesis. In the trial with quantifiable data, random glucose concentration increased by 5±10.2 mg/dl in the rhGH treatment group.35 Since the trial only reported the change from baseline in the rhGH group, weighted mean differences could not be calculated.35 In the other two trials, casual glucose concentrations, interpreted as random, were stated to have remained within the non-diabetic range in all subjects throughout the study.16,34

Fasting Blood Glucose. Fasting blood glucose concentrations were evaluated in three controlled trials.4,24,27 The first trial only reported the change from baseline in the rhGH group, so the weighted mean difference could not be calculated. In that trial, the fasting glucose concentration increased by 7.2±26.1 mg/dl.27 The other two trials were amenable to quantitative synthesis. One of the trials4 evaluated both lower and higher rhGH doses versus placebo while the other compared rhGH to no therapy.24 Upon statistical pooling, there was a significant increase in fasting blood glucose in the rhGH treated group versus control (WMD 5.68 mg/dl, 95 percent CI 0.43 to 10.93 mg/dl). (Figure 24) No statistical heterogeneity was detected. There were too few trials to conduct evaluation of publication bias. The direction of effect was qualitatively the same in both trials. Similarly, both arms of the Schnabel trial showed a similar direction and magnitude of effect suggesting a lack of a dose response effect with rhGH.4

Figure 24. KQ4 — meta-analysis of change from baseline in fasting blood glucose in CF patients treated with rhGH.

Figure 24

KQ4 — meta-analysis of change from baseline in fasting blood glucose in CF patients treated with rhGH. Legend: CF=cystic fibrosis; FBG=fastng blood glucose; rhGH=recombinant human growth hormone Note: The squares represent individual point estimates. (more...)

Fasting blood glucose levels were also reported in a prospective, single arm, observational study that administered rhGH therapy to all participants. There was no significant change from baseline in fasting blood glucose in patients receiving rhGH (on calculation using the data from the observational study, change=5±10 mg/dl).45

Stimulated Blood Glucose. Stimulated blood glucose concentrations were evaluated in two controlled trials.4,27 The first trial only reported the change from baseline in the rhGH group so weighted mean difference could not be calculated. In that trial, the stimulated glucose concentration was reduced by 25.2±149.3 mg/dl versus baseline.27 The other trial evaluated a lower and higher dose of rhGH versus placebo.4 Upon statistical pooling of the lower and higher dose arms versus placebo, there was no significant change in stimulated glucose concentrations with rhGH versus no treatment (WMD 4.92 mg/dl, 95 percent CI −15.12 to 24.98 mg/dl). (Figure 25) There were too few trials to conduct evaluations of publication bias and statistical heterogeneity.

Figure 25. KQ4 — meta-analysis of change from baseline in stimulated blood glucose in CF patients treated with rhGH.

Figure 25

KQ4 — meta-analysis of change from baseline in stimulated blood glucose in CF patients treated with rhGH. Legend: BG=blood glucose; CF=cystic fibrosis; rhGH=recombinant human growth hormone Note: The squares represent individual point estimates. (more...)

Postprandial Blood Glucose. In the one controlled trial where it was evaluated, postprandial blood glucose concentrations were not significantly changed in the rhGH groups versus placebo (elevated by 10 mg/dl).24

Other Blood Glucose Parameters. In one controlled trial evaluating glucose concentrations which were not defined, final blood glucose concentrations were higher in both the rhGH alone arm (97±12 mg/dl) and rhGH plus glutamine arm (95±6 mg/dl) of the trial versus glutamine alone arm (90±6 mg/dl). However, statistical significance is not known since the table within the paper suggests the p-value is less than 0.05 but the text in the results section specifies that it is not significant.30

One single-arm observational study reported no changes in glucose parameters with rhGH therapy versus baseline but did not specify which glucose outcomes were measured.41 Another observational study reported no significant changes in fasting insulin-blood glucose ratio with rhGH treatment in CF patients.45

Glucose Intolerance and Diabetes. The development of glucose intolerance and diabetes were reported in five controlled trials.4,24,26,34,103 (Table 23) No patients developed glucose intolerance or diabetes over the duration of investigation in four of those studies (6 to 12 months).24,26,34,103 One trial did report a hyperglycemic episode in one patient receiving high dose rhGH, but no subjects in the study developed diabetes.4

Glucose intolerance was also reported in three single-arm observational studies.42,43,45 (Table 24) In one study, none of the patients developed hyperglycemia while being treated with rhGH.42 Furthermore, in the one study that monitored rhGH treatment patients for glucosuria, none of the patients had glucose in their urine.45 However, two female patients receiving rhGH developed glucose intolerance during rhGH treatment in a study.43 One patient was initiated on insulin therapy and continued rhGH therapy, while the other patient discontinued rhGH.43 Since there was no information given on confounders, it is difficult to assess causality in these two patients.

The onset of diabetes was reported in a case report where a patient received rhGH therapy for three years.46 While the temporal relationship between rhGH use and diabetes is reasonable, there was no report of improvement in glucose control after dechallenge, no rechallenge, and the patient had an important confounder. The patient had a pancreatic resection right before the onset of diabetes and endocrine function of the pancreas was specifically said to have been preserved until the pancreatic resection. As such, the causality between rhGH therapy and diabetes in this case is weak.46

Injection Site Reactions

Injection site reactions were reported in two observational studies.41,43 One observational study with seven patients reported minor bruising at the injection site in a majority of the subjects (actual number not provided) upon initial treatment, but this effect subsided with increased parental competence in administering rhGH.41 One patient in another observational study reported discomfort secondary to injection, but remained in the study.45 Further information regarding the injection site reactions was not provided in these studies.41,43

Liver Transaminases

The impact of rhGH on liver transaminases was reported in two single-arm observational studies.41,47 Transient elevations in liver transaminase concentrations occurred unexpectedly in two patients receiving rhGH in one study but actual laboratory values and further information was not provided.41 In another study, two patients with CF related liver disease experienced an improvement in liver transaminases after rhGH therapy was initiated; however, one patient’s liver function began to deteriorate again after 2 months of rhGH therapy.47

Study Withdrawals

There were no patient withdrawals in the majority of the controlled trials. (Table 25) One trial reported four withdrawals but did not specify from which treatment arm or the reason why the patients left the study.4 Two patients each from both arms, treatment and no treatment, withdrew in one trial due to fear of injections, commencement of enteral feeds, loss to followup, and development of CF-related diabetes (CFRD).103 Fear of injections was also a reason for one of the two patient withdrawals in another trial; the other withdrawal was due to geographic relocation.24 In another study, a patient left the control group in order to be evaluated for a lung transplant.27

Table 25. Patient withdrawals and reasons from controlled trials evaluating the use of rhGH in CF patients.

Table 25

Patient withdrawals and reasons from controlled trials evaluating the use of rhGH in CF patients.

Discussion

Endogenous growth hormone (GH) regulates the utilization of glucose in many cells throughout the body and impacts insulin sensitivity.118 In general, GH therapy decreases insulin sensitivity and induces a compensatory rise in insulin concentrations.119 However, the impact of rhGH therapy on the development of diabetes remains uncertain. Blethen and colleagues collected data on 47,000 patient-years of rhGH treatment in children without CF and concluded that rhGH is unlikely to cause diabetes unless the patient has preexisting risk factors.120 Another large study with 23,333 children without CF revealed that treatment with rhGH resulted in an increased risk for the development of diabetes.121 Again, these authors suggested that rhGH-induced diabetes developed in predisposed individuals.

Due to the nature of CF, patients are at increased risk for developing impaired glucose tolerance or diabetes, with almost 50% of CF patients developing CF-related diabetes (CFRD).122 CFRD has been primarily attributed to a decrease in insulin secretion; however, a decrease in insulin sensitivity may also play a role.123 The high viscosity of pancreatic secretions is responsible for damage to the beta cells within the pancreas. The loss of beta cells leads to a decrease in insulin production and development of CFRD. The role of insulin resistance in the development of CFRD is still uncertain due to inconsistent findings.124,125 Several studies have shown that the development of CFRD has been linked to a decrease in pulmonary function (both FEV1 and FVC) and trended towards increases in mortality.123,126,127 In one retrospective study, the increase in mortality due to diabetes was primarily manifested in females with CF.128 However, prudent monitoring and early diagnosis along with aggressive treatment have attenuated the difference in mortality between patients with and without CFRD.129 As such, it is important to determine the impact of rhGH therapy on glucose parameters in CF patients in controlled trials.

While numerous trials and studies evaluated the impact of rhGH on glucose parameters, almost all trials compared rhGH effect to no treatment, rather than placebo or another therapy. The increase in fasting glucose concentrations was 6.0 mg/dl in the no treatment group of one trial and 1.2 mg/dl in the placebo group of another. One trial and the observational studies were single arm trials with no control group. As such, a slight increase in glucose concentrations might be related to the duration of the trial and not to the direct impact of rhGH on glucose. This limits the strength of evidence for this adverse event evaluation.

Taken together, it seems that rhGH therapy may slightly increase transient markers of glucose control but does not seem to impact A1c. While glucose parameters were elevated in patients treated with rhGH, it is uncertain if this is truly the effect of treatment or part of the natural progression of CFRD in these patients. Although we showed no significant effect of rhGH on A1c, there has been concern that A1c is an unreliable marker of glucose tolerance in patients with CF and should be interpreted with caution.122,130 Future placebo controlled trials need to stratify patients based on preexisting risk factors to see if rhGH has a significant effect on blood glucose parameters and progression to CFRD.

We had insufficient evidence to evaluate the impact of rhGH on injection site reactions. Some observational studies have reported minor bruising and discomfort at injection sites; however, the effect is limited and resolves with improved administration technique.41,45 Due to the nature of proteins, local reactions at the injection site may theoretically occur.131 The limited occurrence of injection site reactions in non-CF patients, including pediatric patients, is further supported by the information in package inserts of several rhGH products.121–133 Rotation of the injection site is proposed to minimize this adverse effect but this was not evaluated in CF patients.131

We had insufficient evidence to evaluate the impact of rhGH on liver function. There have been reports of unexpected transient elevations in liver transaminases during rhGH therapy41. In the prescribing information for rhGH products, liver enzyme elevation is not listed as a potential side effect.131–135 Furthermore, postmarketing surveillance has not uncovered any liver-related side effects.133

Key Question 5. What is the risk of malignancy associated with rhGH use as determined by: (a) markers of cancer risk with rhGH (IGF-I increases over 100 ng/ml or IGFBP-3 decreases over 1000 ng/ml) from studies of rhGH in people with CF and by (b) assessment of evidence on cancer incidence from non-CF patients receiving modest doses of rhGH (0.2mg/kg/week to 0.6mg/kg/week) for disorders such as growth hormone deficiency (GHD) and idiopathic short stature (ISS)?

Key Points

  • In patients with CF, there appears to be an increase in IGF-I levels in patients treated with rhGH compared to control.
  • There is insufficient evidence to determine the impact of rhGH treatment on IGFBP-3 levels.
  • In patients with GHD or ISS, there is little evidence to evaluate the effects of rhGH treatment on cancer risk.

Detailed Analysis

Study Design and Population Characteristics

The studies that evaluate markers of cancer risk with rhGH use in patients with CF are from the same set of studies used to evaluate Key Question 1 and are summarized in Table 3Table 7. Two epidemiological studies reported the incidence of cancer subsequent to rhGH use in patients with idiopathic GHD or ISS, and there is one case report of cancer.100–102

Outcome Evaluations

Biomarkers in CF Populations

Only one controlled trial reported the change in IGF-I concentrations from baseline, precluding quantitative analysis (Table 26). In that trial, the change in IGF-I concentration from baseline was 188±160 ng/ml in the rhGH group and 31± 117 ng/ml in the no treatment group (p<0.03 for the comparison of values at the end of the study).24,25

Table 26. Biomarkers in controlled trials of CF patients treated with rhGH.

Table 26

Biomarkers in controlled trials of CF patients treated with rhGH.

In three controlled trials, the final IGF-I concentrations identified with rhGH treatment were compared to no treatment (Table 26). In all of those trials, IGF-I concentrations were higher with rhGH therapy than no treatment (p<0.05 for all) with values at least 100 ng/ml higher with rhGH therapy than no treatment.33–35 In another trial, the final IGF-I concentrations were compared between the rhGH group, rhGH plus glutamine group, and glutamine alone group (Table 26).30 IGF-I concentrations in the rhGH group and rhGH plus glutamine group were over 100 ng/ml higher than the glutamine alone group.30 The rhGH alone and rhGH and glutamine groups both had significant increases in IGF-I concentrations from baseline (p<0.05 for each group), while the glutamine alone group did not have a significant increase from baseline (p-value not reported). In a single-arm observational study (n=5), there were significant increases in IGF-I from baseline to 12 months and 24 months (baseline 0.67±0.13 ng/ml; 12 months 1.30±0.38 ng/ml, p<0.01 versus baseline; 24 months 1.86±0.83 ng/ml, p<0.01 versus baseline and p<0.02 versus 12 months).44 Another single-arm observational study (n=9) also found significant increases in IGF-I from baseline (baseline 0.9±0.5 ng/ml; 6 months 3.33±1.9 ng/ml, p=0.003 versus baseline; 12 months 2.6±1.8, p=0.044 versus baseline).45 Neither study experienced increases in IGF-I greater than 100 ng/ml during the study period.44,45

No trials reported changes in IGFBP-3 concentrations from baseline, precluding quantitative analysis. The aforementioned study by Darmaun did report final IGFBP-3 concentrations in the rhGH, rhGH plus glutamine, and glutamine alone groups, where only the rhGH alone group provided significant difference from baseline (p<0.05).30 IGFBP-3 concentrations were qualitatively higher in the rhGH group and rhGH plus glutamine group than the glutamine alone group but this was not statistically evaluated.30 (Table 26) A single-arm observational study (n=5) reported IGFBP-3 levels subsequent to rhGH therapy, showing nonsignificant increases at 12 months and significant increases from baseline at 24 months (baseline 1700±490 ng/ml; 12 months 2200±310 ng/ml, p-value not reported; 24 months 3000±490 ng/ml, p<0.01 versus baseline).44

Malignancy

We identified no controlled rhGH trials evaluating malignancy in CF populations. There was one case report identified which described a CF patient who developed pancreatic cancer.46 Two observational, single-group evaluations from the same registry and a case report evaluated the impact of rhGH on malignancy in non-CF populations with GHD or ISS (Table 27).100–102

Table 27. Evidence of cancer with rhGH therapy in observational studies in GHD and ISS populations.

Table 27

Evidence of cancer with rhGH therapy in observational studies in GHD and ISS populations.

A female with CF underwent bilateral lung transplantation at age 12 years and was treated with the following immunosuppressants: prednisone, cyclosporine A, and mycophenolate mofetil.46 At age 15 years, rhGH 2.2 mg/day was initiated because of growth retardation.46 Before initiating rhGH, IGF-I was 141 ng/ml but did not significantly improve in the 3 years of rhGH therapy, and linear growth remained poor.46 At age 18 years, a pancreatic mass was detected and histology revealed a ductal pancreatic adenocarcinoma, treated with pancreatic resection.46 The occurrence of cancer appeared to have a temporal relationship with the administration of rhGH, but other possible causes of malignancy were not reported. Eight months after pancreatic resection, the patient developed metastatic disease and died.46 Because of the nature of the malignancy, it could not resolve on its own after withdrawal of rhGH and a rechallenge of rhGH therapy was not administered. Based on the presentation of the case, it is possible that the malignancy was related to rhGH therapy but more information would be needed to more clearly determine causality.

The risk of developing leukemia was evaluated in patients from the National Cooperative Growth Study (NCGS) using registry data from 1985 through 1995. Out of the 12,697 patients with either GHD or ISS but without risk factors for cancer, there were two cases of leukemia in 37,772 patient-years of rhGH treatment.100 In a subsequent evaluation of the registry from 1985 through 2006 by NCGS investigators, the occurrence of all types of cancer was evaluated. Out of 33,171 patients with either GHD or ISS but without risk factors for cancer, 29 new-onset malignancy were observed in patients without previous risk factors out of 178,464 years of GH exposure.102 Using background rates of cancer in an age-adjusted general population, Bell and colleagues calculated a standard incidence ratio (SIR) and 95 percent CI, which represent the number of observed cases over the number of expected cases.102 No difference was seen between patients exposed to rhGH and the general population (SIR 1.12, 95 percent CI 0.75 to 1.61).102 These observational studies are limited because the doses of rhGH utilized and the manufacturers are not reported and because cancers occurring after rhGH discontinuation would not be captured.

In 2,712 patients with cancer risk factors (which includes prior malignancy, radiation exposure, bone marrow transplant, chemotherapy, neurofibromatosis, immunosuppressant use, and certain chromosomal disorders like Down syndrome), there were eight cases of leukemia in 8,962 patient years of rhGH treatment in the original NCGS study.100 In 2,500 patients with history of malignancy, 49 cases of second neoplasms among 10,750 years of exposure occurred in the followup NCGS study.102 The most commonly identified cancer was leukemia, with 18 cases being observed.102

In a 9.7 year old male with short stature, rhGH was initiated to treat GHD.101 The patient underwent a cranial MRI at age 13.4 years due to persistent headaches which revealed a large suprasellar mass. The patient was eventually diagnosed with intracranial suprasellar choriocarcinoma.101 The occurrence of cancer appeared to have a temporal relationship with the administration of rhGH, but other possible causes of malignancy were not reported. Therapy with rhGH was discontinued, and the patient was successfully treated with radiation and chemotherapy.101 Because of the nature of the malignancy, it could not resolve on its own after withdrawal of rhGH and a rechallenge of rhGH therapy was not administered. Based on the presentation of the case, the causality of malignancy from rhGH therapy is possible.

Discussion

According to the Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute, patients with CF have a higher incidence of digestive cancers including cancers of the esophagus, intestine, liver, biliary tract, or pancreas but no increased rates of any other type of cancer.134–136 As patients with CF are continuing to live longer, the potential risks that rhGH may have on cancer are important to evaluate.

It is hypothesized that IGF-I induces cellular proliferation and angiogenesis or inhibits apoptosis while IGFBP-3 inhibits IGF-I induced proliferation and may decrease migration and adhesion.137,138 Data in non-CF patients suggests that IGF-I and IGFBP-3 concentrations are associated with the development of certain types of cancer. In a prospective case-control study from the Physicians’ Health Study, increases in IGF-I by 100 ng/ml were associated with a 1.69 relative risk of developing colorectal cancer (95 percent CI 1.07 to 2.67), while increases in IGFBP-3 by 1000 ng/ml corresponded with a 0.54 relative risk of developing cancer (95 percent CI 0.34 to 0.84).139 Another case-control evaluation in patients with breast cancer from the Nurses’ Health study found no relationship between IGF-I or IGFBP-3 with either premenopausal or postmenopausal breast cancer.140 When evaluating patients stratified by quartiles of IGF-I levels, patients with highest IGF-I levels had no difference in the risk of breast cancer compared with patients with the lowest IGF-I levels (RR 1.00, 95 percent CI 0.73 to 1.37).140 Similar results were seen when comparing the highest and lowest IGFBP-3 quartiles (RR 1.07, 95 percent CI 0.79 to 1.45).140

In a meta-analysis of cohort and case-control studies, elevated IGF-I concentrations were associated with increased risk of prostate (OR 1.83, 95 percent CI 1.03 to 3.26), colorectal (OR 1.58, 95 percent CI 1.11 to 2.27), and premenopausal breast cancer (OR 1.93, 95 percent CI 1.38 to 2.69) but not associated with postmenopausal breast (OR 0.95, 95 percent CI 0.62 to 1.33) or lung cancer (OR 1.01, 95 percent CI 0.49 to 2.11).141 Elevated concentrations of IGFBP-3 were also associated with an increased risk of premenopausal breast cancer (OR 1.96, 95 percent CI 1.28 to 2.99); however, there was no effect on the rates of prostate (OR 0.88, 95 percent CI 0.61 to 1.28), colorectal (OR 0.77, 95 percent CI 0.38 to 1.66), postmenopausal breast (OR 0.97, 95 percent CI 0.53 to 1.77), or lung cancer (OR 0.83, 95 percent CI 0.38 to 1.84).141

In our comparative effectiveness review, rhGH therapy in CF patients caused higher on-treatment IGF-I values than control patients and differences between the groups exceeded 100 ng/ml. Interpretation of elevations in IGF-I should be done cautiously as CF patients suffer from IGF-I deficiency compared to healthy populations.142 While data is limited to a single small RCT, IGFBP-3 concentrations were higher in the rhGH group than the glutamine group but the absolute difference was only 0.89 ng/ml and far away from difference of 1,000 ng/ml. Extrapolating from prospective case control data in non-CF populations, the increases in IGF-I seen with rhGH therapy in this trial may be clinically relevant but the relevance of slight IGFBP-3 elevations cannot be determined. More research is required to determine if rhGH therapy may increase IGF-I and IGFBP-3 levels above the normal reference range and if that is a marker of increased malignancy risk.

Key Question 6. In patients with CF, how is efficacy, effectiveness, safety or adverse events impacted by rhGH dose, therapy duration, baseline nutritional status, and concurrent medical therapies?

Key Points

  • Only one trial provides insight into the dose response nature of rhGH in patients with CF. In this trial, no significant differences were seen between the higher and the lower dose groups for any evaluated parameter.
  • Several trials allow the evaluation of the duration of rhGH therapy and outcomes, suggesting that greater benefits are derived from longer-term therapy but that hyperglycemia is also more common. These are qualitative differences however, and we cannot be sure that they are not due to chance.
    • Trials with 1 year of rhGH therapy significantly increased percent predicted FVC, absolute FEV1, and height compared to control, while 6 months of rhGH therapy showed no effect.
    • Trials with 1 year of rhGH therapy significantly increased fasting glucose concentrations, while trials of 6 months duration showed no effect.
  • rhGH has not been studied in patients with CF who have nutritional deficiencies that are not being addressed with enteral nutrition. We cannot determine the benefits of rhGH therapy in patients with unaddressed nutritional deficiencies.
  • The usage of concurrent medical therapies in patients enrolled in trials evaluating rhGH therapy was sparingly reported, so the differential effect on rhGH efficacy could not be assessed.

Detailed Analysis

Study Design and Population Characteristics

Studies to answer Key Question 6 are derived from the same set of studies used to evaluate Key Question 1 and are summarized in Table 3Table 5.

Outcome Evaluations

Dose

All controlled trials except one used very similar doses of rhGH. Seven trials used 0.3 mg/kg/week doses,16,24,27,30,33,35,39 another trial used 0.27 to 0.35 mg/kg/week doses,26 and a third trial used 0.3 to 0.35 mg/kg/week doses.34 None of these trials reported results in patients receiving different doses and does not provide insight into the impact of dose on outcomes. The exception is a three-arm trial with lower dose, a higher dose, and a placebo arm which will be used to assess the impact of dose on efficacy and safety.4 Since the two rhGH arms in the trial only have 42 patients between them, it is underpowered and only qualitative insight can be garnered.

Efficacy. One randomized controlled trial evaluated two doses of rhGH therapy in CF patients, with patients being assigned to 0.273 mg/kg/week (n=22), a dose slightly lower than the other trials, or 0.49 mg/kg/week (n=20), or placebo (n=21).4 Since p-values were not provided for the higher versus lower comparison, we compared the changes from baseline in the lower and higher groups using an unpaired Student t-test and calculated the p-value from the provided data using the Primer of Biostatistics (McGraw-Hill, Stanton, CA).

There were no qualitative differences between the higher and lower dose groups in FEV1 Z-score (change from baseline −0.04±0.30 versus −0.03±0.32, p-value not reported, our calculated p=0.918).4 There were also no statistically significant differences between the higher dose group than the lower dose group in percent predicted FVC (change from baseline 6.0±11.2 versus 3.1±13.1, p-value not reported, our calculated p=0.447) and percent predicted FEV1 (change from baseline 4.3±13.4 versus 3.5±12.3, p-value not reported, our calculated p=0.841).4

The higher dose group was no different than the lower dose group in both height velocity (change from baseline 6.8±4.3 versus 5.6±2.9 cm/year, p-value not reported, our calculated p=0.291) and height velocity Z-score (change from baseline: higher dose 2.6±2.7; lower dose 1.5±2.6, p-value not reported, our calculated p=0.186).4

The higher and lower dose groups experienced similar increases in weight (change from baseline 2.2±2.3 kg versus 2.4±1.9 kg, p-value not reported, our calculated p=0.759.), LBM (change from baseline 2.3±2.5 kg versus 2.5±2.4 kg, p-value not reported, our calculated p=0.793), and BMI Z-score (change from baseline 0.1±0.6 versus 0.0±0.6, p-value not reported, our calculated p=0.593).4

Safety. There were no significant differences between the higher dose group and the lower dose group in IGF-I Z-scores (1.66±1.53 versus 1.00±1.06, p-value not reported, our calculated p=0.109) and IGFBP-3 Z-scores (0.66±1.41 versus 0.5±1.6, p-value not reported, our calculated p=0.734).4

Duration

Seven controlled trials had a duration of therapy of 1 year,16,24,27,33–35,39 two trials had a 6 month duration,4,109 and a final trial had 1 month of followup.30 Trials were amenable to subgroup analysis based on duration of followup on the same outcomes meta-analyzed in Key Questions 1, 2, and 4. Results of subgroup analysis are presented in Table 29.

Table 29. Subgroup analyses based on duration of rhGH therapy.

Table 29

Subgroup analyses based on duration of rhGH therapy.

Efficacy. All controlled trials reporting absolute FVC were 1 year in duration precluding evaluation for this key question. There was no statistically significant difference in percent predicted FVC and absolute FEV1 between patients treated for 1 year compared to those treated for 6 months. (Table 29) However, trials of 1 year in duration showed statistically significant improvements with rhGH therapy compared to control in percent predicted FVC and absolute FEV1, while trials of 6 months duration showed no difference.

There were no statistically significant differences between 1 year of treatment compared with 6 months of treatment in absolute height and height velocity. (Table 29) Height Z-score was only reported in trials of 1 year duration. One year of rhGH therapy led to statistically significant increases in height compared to control, while 6 months of rhGH therapy was not significantly different from control. There were no improvements in weight outcomes with 1 year of treatment compared with 6 month in absolute weight, and lean body mass. (Table 29) Weight velocity, weight Z-score, BMI, and percent IBW were only reported in trials of 1 year duration. BMI Z-score was only reported in trials of 6 months duration. The differential effects of treatment duration on BMC could not be assessed because all trials were 1 year in duration. The differential effects of treatment duration on hospitalization rate could not be assessed because all trials were 1 year in duration.

In an open label extension trial, patients were given rhGH therapy for a second year, regardless of the treatment they were allocated to in the original 1 year clinical trial.35 Those patients continuing rhGH treatment had similar effects over the second year of study as those newly initiating rhGH treatment in height velocity (5.9±2.1 versus 6.2±1.2 cm/year; p-value not reported) and change in LBM from the initiation of open-label rhGH (3.9±1.4 versus 4.1±2.0 kg, p-value not reported).35 Weight velocity was qualitatively higher in those continuing rhGH as compared to those initiating rhGH (6.0±1.7 versus 4.6±3.1 kg/year, p-value not reported). Those continuing rhGH had similar pulmonary effects as those initiating rhGH in the change from baseline for absolute FVC (1.1±0.2 versus 1.6±0.1 L, p-value not reported), absolute FEV1 (0.6±0.3 versus 0.5±0.4 L, p-value not reported), and BMC (177±69 versus 163±75 g, p-value not reported).35 There was a qualitatively greater hospitalization rate in those continuing rhGH than those initiating rhGH (2.1±2.1 versus 0.8±0.4 events per year, p-value not reported).35

Safety. Upon subgroup analysis based upon rhGH treatment duration, the fasting blood glucose had no significantly different changes from baseline in the rhGH group compared to control groups in patients treated for 1 year versus those treated for 6 months. (Table 29) One year of rhGH therapy significantly increased fasting blood glucose compared to control, but 6 months of rhGH therapy had no effect on fasting blood glucose compared to control. The differential effects of treatment duration could not be assessed for A1c because all trials were for 1 year, nor assessed for stimulated blood glucose because all trials were for 6 months.

Glucose parameters in the second year of rhGH treatment in the trial by Hardin and colleagues in 2005 were not reported.35 After the second year of therapy, IGF-I levels increased in those initiating rhGH therapy (from 125±27 ng/ml to 324±29 ng/ml, p-value not reported) and did not significantly change from the initiation of open-label rhGH in those continuing rhGH therapy (from 286±91 ng/ml to 319±25 ng/ml, p-value not reported).35

Baseline Nutritional Status

The baseline nutritional status in patients enrolled in controlled trials evaluating rhGH therapy was sparingly reported. (Table 28) The nutritional status was not specified in five trials.26,27,30,33,39 The five remaining trials reported specific nutrition-related inclusion and exclusion criteria, which allude to the nutritional status of patients: three trials specified that patients must have good or adequate caloric intake,4,24,34 one trial excluded patients who required parenteral caloric supplementation,16 and one trial included only patients who required and received enteral nutrition.35 No trial reported that patients had poor nutrition precluding comprehensive subgroup analysis. We focused on the trial where patients required and received enteral nutrition in this portion of the key question since it addresses, in part, the concern about whether rhGH would have effects over and above improving nutritional status in nutritionally at risk individuals.

Table 28. Concurrent therapies and nutrition in controlled trials evaluating rhGH.

Table 28

Concurrent therapies and nutrition in controlled trials evaluating rhGH.

Efficacy. One randomized controlled trial by Hardin and colleagues evaluated patients who had been receiving enteral nutrition overnight via gastronomy tube for at least 2 years prior to study enrollment.35 Patients treated with growth hormone (n=9) showed statistically significant improvements over patients in the control group (n=9) during 12 months of therapy.35 There were statistically significant improvements in pulmonary function as measured by absolute FVC (p<0.05 between groups at 12 months) and absolute FEV1 (p<0.05 between groups at 12 months).35 Significant improvements with rhGH therapy were also seen in anthropometric measures such as height velocity, height Z-score, weight velocity, weight Z-score (p<0.05 between groups for all endpoints at 12 months), BMI and LBM (p<0.05 between groups for change from baseline in endpoints at 12 months).35 Benefits were seen with increases in BMC over 12 months of therapy as well (rhGH group 176±22 g/year; control group 34±15 g/year, p=0.02).35 The rate of hospitalizations was also significantly fewer in patients receiving rhGH compared to patients with no therapy (rhGH group 1.1±1.0 hospitalizations/year; control group 3.0±2.0 hospitalizations/year, p=0.003).35 The results of this trial are similar to the effects seen in the overall set of trials evaluating the use of rhGH in CF patients, suggesting that the effect of rhGH is applicable to patients who require and are receiving enteral nutrition.

Safety. In patients receiving enteral nutrition for 1 year, there was no significant change from baseline in casual blood glucose in those treated with rhGH (baseline 87±11 mg/dl; 1 year 92±9 mg/dl, p=not significant).35 Changes from baseline in casual blood glucose in the control group were not reported.35 There were significant increases in IGF-I in the rhGH group compared to baseline after 1 year, but no changes in the control group (baseline for both groups 119±42 ng/ml; rhGH at 1 year 286±91; control at 1 year 125±27; p-values not reported).35

Concurrent Medical Therapies

The usage of concurrent medical therapies in patients enrolled in trials evaluating rhGH therapy was sparingly reported. (Table 28) Therefore, the effect that concurrent therapy may have on rhGH efficacy and safety could not be assessed.

Discussion

Evidence for a dose-response relationship is poor. Further study is required to establish efficacy of rhGH based upon dose. From the single trial which evaluated two different doses of rhGH, there were mixed results on the effect of a higher dose compared to a lower dose.4 Pulmonary function as measured by percent predicted FVC and percent predicted FEV1 trended towards greater improvement in the higher dose group than the lower dose group.4 There were slightly greater improvements in height parameters in the higher dose group but similar effects on weight outcomes.4 None of these comparisons were statistically significant, and would require a larger sample size to determine the true differential effects of the two doses if any exist.

In general, controlled trials conducted for 1 year exhibited greater rhGH efficacy compared to non-active control than those conducted for 6 months but greater increases in serum glucose occur. Trials which continued rhGH therapy for a second year continued to show improvement in intermediate outcomes. None of these comparisons were statistically significant and would require a larger sample size to determine the true differential effects of the two doses if any exist.

With regard to baseline nutritional status, none of the trials specified that they had patients with poor nutrition. There are no data available to assess the efficacy of rhGH in patients with inadequate nutrition. Therefore, consideration of rhGH therapy should occur after a patient is receiving adequate caloric intake. One trial evaluated patients who all received enteral nutrition, and showed that there is efficacy of rhGH in addition to enteral nutrition. No trials evaluated rhGH use in addition to parenteral nutrition, so its efficacy in this clinical scenario is uncertain. In addition, there were no studies in this report comparing rhGH therapy to a strategy where additional caloric consumption was provided.

The underreporting of concurrent medical therapies in patients precluded analysis or discussion on the benefit of rhGH in addition to specific therapies.

Key Question 7. In patients with CF, how do the efficacy, effectiveness, safety or adverse events of treatment with rhGH differ between subgroups of patients? Subgroup characteristics of interest include, but are not limited to: age (pre-pubertal, pubertal, post-pubertal); gender; baseline clinical status (height, weight, lean body mass, pulmonary function, exercise tolerance, nutritional status); and/or the nature, extent, and effectiveness of prior treatment

Key Points

  • The age of the patient may impact rhGH efficacy.
    • In an individual patient data merged analysis of trials, both prepubertal and adolescent patients had significant improvements in height, weight, lean body mass, and hospitalizations as compared to their respective control populations. Prepubertal patients receiving rhGH did not have significant increases in FEV1 and the percent predicted FEV1 was significantly lower than prepubertal control patients. In contrast, adolescent patients receiving rhGH had significant improvements in FEV1 and percent predicted FEV1.
    • Prepubertal patients receiving rhGH seem to derive greater benefits on height than pubertal patients receiving rhGH but lesser benefits on weight, BMI, and percent ideal body weight. Pubertal patients receiving rhGH also seem to derive greater increases in absolute FVC, FEV1, and bone mineral content but fewer hospitalizations and smaller increases in percent predicted FVC than prepubertal patients.

      Since these observations are derived from comparing the results of a pooled analysis from trials consisting of only prepubertal patients to those with only pubertal patients, these results are only hypothesis generating.

  • While most controlled trials were conducted predominantly in males, the impact of gender on outcomes of rhGH therapy could be qualitatively assessed in one pooled analysis. The authors of the analysis did not report p-values or whether the comparisons were statistically significant and did not provide patient numbers precluding our ability to calculate these p-values.
    • In prepubertal patients not receiving rhGH therapy, no difference in height velocity occurred between the genders the year before treatment allocation but females had greater weight velocity. In pubertal patients not receiving rhGH therapy, females had greater height and weight velocity than males the year before treatment allocation.
    • In prepubertal patients, the first 6 months of rhGH therapy provided similar increases in height and weight velocity between genders but in months 6 to 12, females had greater height velocity while males had greater weight velocity.
    • In pubertal patients, the first 6 months of rhGH therapy provided similar increases in height velocity between genders but females had greater increases in weight velocity. In months 6 to 12, females had greater height and weight velocities than males.
    • The occurrence of adverse effects associated with rhGH therapy in males and females was not individually determined.
  • The impact of baseline clinical status on rhGHs clinical outcomes was assessed in two trials. In the first trial, those with a baseline height Z-score below −2.2 had a similar increase in height Z-score on rhGH therapy. In the second trial, a higher baseline percent predicted FEV1 was positively correlated with the change of weight associated with rhGH therapy. The occurrence of adverse events associated with rhGH therapy in patients with different baseline clinical status could not be determined.

Detailed Analysis

Study Design and Population Characteristics

Studies to answer Key Question 7 are derived from the same set of studies used to evaluate Key Question 1 and are summarized in Table 3Table 5.

Outcome Evaluations

Age

Six controlled trials specifically evaluated prepubertal patients,16,24,30,33,35,39 one trial evaluated pubertal adolescents exclusively,34 and three trials did not specify the pubertal status of patients.4,27,109

Efficacy. We pooled together controlled trials that evaluated only prepubertal patients and then qualitatively compared the magnitude of rhGH’s effect versus control to that derived from the single trial which evaluated only pubertal patients. (Table 30) Although statistical comparison of the two subgroups was not conducted, the results can be compared with the understanding that this comparison is hypothesis generating only. Patients who have reached puberty had qualitatively greater response to rhGH therapy in absolute FVC and absolute FEV1 compared to prepubertal patients. (Table 30) However, prepubertal patients had greater improvement in percent predicted FVC than pubertal patients, and the differential effect on percent predicted FEV1 could not be assessed.

Table 30. Subgroup analyses based on pubertal status of patients enrolled.

Table 30

Subgroup analyses based on pubertal status of patients enrolled.

For absolute changes in height, the prepubertal patients experienced greater response than pubertal patients. The effect that pubertal status may have on response to rhGH on height velocity and height Z-score could not be assessed. In contrast, pubertal patients experienced greater weight gain, BMI, and IBW with rhGH than prepubertal patients. (Table 30) Weight velocity, weight Z-score, BMI Z-score and LBM in prepubertal and pubertal patients on rhGH could not be assessed.

Pubertal patients accumulated greater BMC than prepubertal patients, likely due to the presence of mature sex hormones, which may aid in the response to rhGH. Exercise tolerance was only assessed in trials enrolling a mix of pubertal status and was not compared between subgroups. Hospitalization rate was reduced more in pubertal patients treated with rhGH than prepubertal patients. (Table 30)

In a meta-analysis by Hardin and colleagues presented as a poster in 2009, results were reported separately for prepubertal children and adolescents.143 At 1 year of rhGH therapy, prepubertal patients in the rhGH group (n=87) experienced significantly better outcomes than prepubertal children in the control group (n=60) in height (138.7±11.7 cm versus 130.1±14.9 cm, p-value not reported but said to be statistically significant), weight (30.3±5.9 kg versus 26.2±6.8 kg, p-value not reported but said to be statistically significant), and LBM (25.8±5.7 kg versus 20.4±3.4 kg, p-value not reported but said to be statistically significant).143 There were no significant differences in absolute FEV1 (1.43±1.5 L versus 1.74±2.7 L, p-value not reported), although the rhGH group had significantly worse percent predicted FEV1 (73.6±24.8 versus 77.0±26.6, p-value not reported but said to be statistically significant).143 The rhGH group also had significantly fewer hospitalizations than the control group (0.55±1.1 versus 1.2±0.9, units of measure not reported, p-value not reported but said to be statistically significant).143

Adolescent patients treated with rhGH (n=54) also had better outcomes than adolescent patients in the control group (n=22) in regards to height (158.2±8.8 cm versus 153.2±7.7 cm, p-value not reported but said to be statistically significant), weight (42.4±7.9 kg versus 39.1±6.5 kg, p-value not reported but said to be statistically significant), and LBM (36.9±6.6 kg versus 33.1±5.9 kg, p-value not reported but said to be statistically significant).143 In contrast to prepubertal patients on rhGH, adolescent patients treated with rhGH experienced significantly better effects than control patients on both absolute FEV1 (2.63±1.27 L versus 1.99±0.77 L, p-value not reported but said to be statistically significant) and percent predicted FEV1 (80.4±23.8 versus 61.7±26.9, p-value not reported but said to be statistically significant).143 Adolescent patients treated with rhGH also experienced significantly fewer hospitalizations than control (0.84±0.84 versus 1.9±1.4, units of measure not reported, p-value not reported but said to be statistically significant).143

The controlled trial by Hardin and colleagues in 2005, evaluating rhGH use exclusively in adolescent patients, included a report of anthropometric results divided by Tanner stage.34 Patients in Tanner stage 3 treated with rhGH (n=6) experienced significantly better outcomes after 1 year of therapy than those without treatment (n=7) in height Z-score (−1.58 versus −3.01, p<0.002), weight Z-score (−1.89 versus −2.34, p<0.002), height velocity (8.3 cm/year versus 4.5 cm/year), weight velocity (7.3 kg/year versus 1.4 kg/year, p<0.002), and BMI (17.5 kg/m2 versus 15.9 kg/m2, p<0.002).34 Patients in Tanner stage 4 treated with rhGH (n=7) also experienced significantly better outcomes after one year of therapy than those without treatment (n=5) in height Z-score (−1.19 versus −2.73, p<0.002), weight Z-score (−1.21 versus −1.54, p<0.002), height velocity (8.5 cm/year versus 5.7 cm/year, p<0.002), weight velocity (8.6 kg/year versus 4.7 kg/year, p<0.002), and BMI (18.7 kg/m2 versus 15.8 kg/m2, p<0.002).34 Measures of variance surrounding these mean values were not reported.34

In the pooled study by Vanderwel and Hardin, three previous controlled trials16,24,34 were combined to evaluate patients with CF in four mutually exclusive subgroups: prepubertal females, pubertal females, prepubertal males, and pubertal males.38 Pubertal status did not appear to affect the response to rhGH on height velocity in female patients. Prepubertal females who did not receive therapy in 1 year (number of patients not reported) showed similar height velocity (5.7±2.4 cm/year) to pubertal females who did not receive rhGH therapy (number of patients not reported, 4.5±1.0 cm/year).38 In the first 6 months of rhGH therapy, prepubertal females who received therapy showed similar height velocity to pubertal females (8.5±2.6 cm/year versus 8.5±1.0 cm/year, p-value not reported).38 During months 6 to 12 of therapy, there were also similar height velocities between prepubertal and pubertal females (7.7±2.7 cm/year versus 8.2±1.3 cm/year, p-value not reported);38 however, there appeared to be differential effects of pubertal status on weight velocity. Nontreated prepubertal females showed a weight velocity of 3.7±2.4 kg/year after 1 year, while nontreated pubertal females showed a weight velocity of 4.0±3.2 kg/year, p-value not reported.38 In the first 6 months of rhGH therapy, prepubertal females experienced similar weight velocity to pubertal females (4.8±3.9 kg/year versus 5.5±3.1 kg/year, p-value not reported).38 During months 6 to 12 of therapy, prepubertal females showed qualitatively lesser weight velocity than pubertal females (2.2±1.8 kg/year versus 6.4±4.6 kg/year).38

In the males evaluated with regard to pubertal status, there did not appear to be any differential effects of rhGH therapy on height velocity, although there were differences between groups in those who did not receive therapy. Prepubertal males who did not receive therapy (number of patients not reported) showed height velocity 5.1±1.0 cm/year in 1 year compared to nontreated pubertal males (number of patients not reported) showing height velocity 2.7±0.2 cm/year, p-value not reported.38 In the first 6 months of rhGH therapy, prepubertal males experienced similar effects on height velocity as pubertal males (8.3±2.4 cm/year versus 8.2±3.6 cm/year, p-value not reported).38 During months 6 to 12 of therapy, there were also similar results between prepubertal and pubertal males (6.8±2.6 cm/year versus 7.0±3.6 cm/year, p-value not reported).38 Similar effects between prepubertal and pubertal males were also seen in weight velocity, although there were differences between groups in those who did not receive therapy. In the nontreated prepubertal males, weight velocity in 1 year was 1.9±1.4 kg/year compared to 3.0±0.4 kg/year in pubertal males, p-value not reported.38 In the first 6 months of rhGH therapy, prepubertal males showed weight velocity 4.1±2.2 kg/year while pubertal males showed 3.3±2.0 kg/year, p-value not reported.38 During months 6 to 12 of rhGH therapy, prepubertal males showed similar weight velocity than pubertal males (3.8±2.7 kg/year versus 5.0±3.0 kg/year, p-value not reported).38 P-values were not reported for these comparisons; we could not conduct our own comparison using unpaired t-tests because the sample size for each treatment group within the pubertal and prepubertal subgroups were not reported.

In the trial by Schnabel and colleagues, the change from baseline in height was negatively correlated with chronological age (r=0.61, p<0.0001).4

Safety. Upon subgroup analysis, there were no qualitative differences between prepubertal and pubertal patients in A1c response to rhGH therapy. Differential effects that pubertal status may play on fasting and stimulated blood glucose could not be assessed.

Trials that reported results in subgroups based upon pubertal status did not report on safety parameters.

Gender

All controlled trials included patients of which more than half were male, precluding subgroup comparisons of trials based on predominant gender. In the pooled study by Vanderwel and Hardin, three previous trials16,24,34 were combined to evaluate patients with CF in four mutually exclusive subgroups: prepubertal females, pubertal females, prepubertal males, and pubertal males.38

Efficacy. In the study by Vanderwel and Hardin, the height velocity in prepubertal females and prepubertal males who did not receive rhGH therapy (number of patients not reported) were similar in the year before trial initiation (5.7±2.4 cm/year versus 5.1±1.0 cm/year, p-value not reported).38 In the first 6 months of rhGH therapy, prepubertal females responded similarly to prepubertal males in height velocity (8.5±2.6 cm/year versus 8.3±2.4 cm/year, p-value not reported).38 During months 6 to 12 of rhGH therapy, height velocity was qualitatively higher in prepubertal females and prepubertal males (7.7±2.7 cm/year versus 6.8±2.6 cm/year, p-value not reported).38 Weight velocity in 1 year was qualitatively higher in nontreated prepubertal females than nontreated prepubertal males (3.7±2.4 kg/year versus 1.9±1.4 kg/year, p-value not reported).38 In the first 6 months of rhGH therapy, weight velocity was similar between prepubertal males and females (4.8±3.9 kg/year versus 4.1±2.2 kg/year, p-value not reported).38 During months 6 to 12 of rhGH therapy, there was a qualitatively lesser weight velocity in prepubertal females than prepubertal males (2.2±1.8 kg/year versus 3.8±2.7 kg/year, p-value not reported).38

Pubertal females tended to have qualitatively greater height velocity and weight velocity than pubertal males (number of patients not reported).38 In the year before trial initiation, greater height velocity occurred in pubertal females than pubertal males (4.5±1.0 cm/year versus 2.7±0.2 cm/year, p-value not reported).38 In the first 6 months of rhGH therapy, prepubertal females showed similar height velocity to prepubertal males (8.5±1.0 cm/year versus 8.2±3.6 cm/year, p-value not reported).38 During months 6 to 12 of rhGH therapy, pubertal females showed qualitatively greater height velocity than pubertal males (8.2±1.3 cm/year versus 7.0±3.6 cm/year, p-value not reported).38 Weight velocity in the year before trial initiation was also greater in pubertal females than pubertal males (4.0±3.2 kg/year versus 3.0±0.4 kg/year, p-value not reported).38 In the first 6 months of rhGH therapy, pubertal females showed greater weight velocity than pubertal males (5.5±3.1 kg/year versus 3.3±2.0 kg/year, p-value not reported).38 During months 6 to 12 of rhGH therapy, pubertal females continued to show greater weight velocity than pubertal males (6.4±4.6 kg/year versus 5.0±3.0 kg/year, p-value not reported).38

P-values were not reported for these comparisons; we could not conduct our own comparison using unpaired t-tests because the sample size for each treatment group within the pubertal and prepubertal subgroups were not reported.

Safety. The study by Vanderwel and Hardin did not report results on safety parameters.38

Baseline Clinical Status

Two trials reported results based on differences in baseline clinical status.4,16

Efficacy. One trial conducted a planned subgroup analysis on the effect of rhGH on height outcomes with regard to baseline height (height Z-score <−2.2, n=9 versus height Z-score >−1.2, n=9).16 Both subgroups experienced similar response to rhGH therapy, with increases of 0.42±0.13 and 0.47±0.4 in height Z-score from baseline, respectively (p=0.3).16 Results for the subgroups in the control group were not reported.16

In the trial by Schnabel and colleagues, the change in weight in percentage from baseline was positively correlated with the baseline percent predicted FEV1 (r=0.61, p<0.0001).4

Safety. No controlled trials reported effects of rhGH therapy on safety parameters, differentiating by baseline clinical status.

Prior Medical Therapy

No controlled trials reported prior medical therapies used and their potential impact on the efficacy of rhGH therapy.

Discussion

The results shown in subgroup analysis and in individual trials suggest that there is benefit to using rhGH therapy across all age groups. However, the magnitude of efficacy appears to differ between age groups.

Upon subgroup analysis, we found that prepubertal patients experienced greater rhGH benefit than pubertal patients in percent predicted FVC and absolute height. For all other outcomes, pubertal patients experienced qualitatively greater benefit to rhGH therapy. It is expected that the pubertal patients experienced less height gain than pubertal patients because they are likely closer to their maximal height.

One interesting finding in the meta-analysis conducted by Hardin and colleagues was that while pulmonary function did not significantly improve with rhGH therapy in prepubertal patients, it did significantly improve with rhGH therapy in adolescent patients.143 It is possible that adolescent patients may not experience the dramatic changes in linear growth that prepubertal patients might, and therefore show improvements in pulmonary function independent of height. It would be most beneficial to know the changes from baseline in all of these parameters rather than the final results at the end of therapy, but the data are only available in an abstract form at this time. We look forward to the publication of the full manuscript of this analysis to elucidate the relationship that age or pubertal status may have on response to rhGH therapy.

When evaluating only Tanner stage 3 or Tanner stage 4 patients,34 both groups of patients had significantly better outcomes with rhGH treatment than without in anthropometric outcomes. However, instead of the comparison of final values as reported, it would be more insightful to account for the variations in baseline status in these two pubertal groups. It would be interesting to see if the magnitude of rhGH effect varies based upon the pubertal status of the patient, rather than simply knowing that rhGH has an effect compared to control. This would be able to give clinicians insight on the value of adding rhGH to a post-pubertal patient’s regimen.

In the study by Vanderwel and Hardin, similar effects on height and weight velocities were seen between prepubertal females and pubertal females and between prepubertal males and pubertal males.38 Schnabel and colleagues found a negative correlation between height gain and chronological age.4 The negative correlation in height gain from baseline and chronological age is likely due the attainment of maximal height.

There is little evidence to determine the impact of gender on rhGH efficacy. In the study by Vanderwel and Hardin, prepubertal females had qualitatively greater response in height velocity than prepubertal males, while prepubertal males had greater weight velocity than prepubertal females.38 In contrast, pubertal females had both greater height velocity and weight velocity than pubertal males.38 These results must be interpreted with caution because it does not account for the height and weight velocities seen in the control groups. In order to make a strong comparison between males and females, we must first determine the mean differences in effect between the rhGH and control groups to find what additional benefit rhGH therapy would have to standard therapy. Ideally, we should be comparing the mean differences in females to the mean differences in males in order to judge the comparative efficacy of the treatment. Unfortunately, due to the underreporting of the sample size of each treatment group within each of the subgroups, we could not calculate the mean differences.

One trial reported results of subgroup analyses on the tallest and shortest patients in the study, finding similar changes in height Z-score from baseline in rhGH-treated patients in either subgroup.16 Results of subgroup analysis in the control group were not reported,16 but this information would have been valuable to determine if the tallest and shortest patients experience different innate changes in height Z-score without therapy over 1 year. Without data in the control group, it is difficult to determine the effect of rhGH in addition to standard therapy.

Schnabel and colleagues found a positive correlation between weight gain and baseline FEV1,4 suggesting that patients with better pulmonary function at baseline have greater weight response to rhGH therapy. Potential mechanisms for this relationship are unclear.

Cover of Effectiveness of Recombinant Human Growth Hormone (rhGH) in the Treatment of Patients With Cystic Fibrosis
Effectiveness of Recombinant Human Growth Hormone (rhGH) in the Treatment of Patients With Cystic Fibrosis [Internet].
Comparative Effectiveness Reviews, No. 23.
Phung OJ, Coleman CI, Baker EL, et al.

AHRQ (US Agency for Healthcare Research and Quality)

In this Page

PubMed Health Blog...

read all...

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...