A study of how growth hormone treatment affects health outcomes in patients with noncancerous craniopharyngioma tumors

Lead Investigator: Odelia Cooper, Cedars-Sinai Medical Center
Title of Proposal Research: A study of how growth hormone treatment affects health outcomes in patients with noncancerous craniopharyngioma tumors
Vivli Data Request: 6532
Funding Source: None
Potential Conflicts of Interest: None

Summary of the Proposed Research:

Craniopharyngioma (CP) is a rare type of tumor that grows near the pituitary gland at the base of the brain. It is estimated that 400 people are diagnosed with this condition in the United States each year. Patients with these tumors are often unable to control their appetites. They gain a lot of weight and develop diseases related to extreme weight gain that make them ill. Some patients have diabetes mellitus, some have fatty liver disease, and some have heart disease. If these diseases are not treated early, they can shorten a person’s lifespan.

Approximately 80% of adults with craniopharyngioma produce too little growth hormone from the pituitary gland. This likely occurs because the tumor presses against the pituitary gland. We believe that treating these patients with growth hormone will prevent or improve outcomes from the diseases caused by the extreme weight gain.

The Hypopituitary Control and Complications Study (HypoCSS) contains information on 9,686 adults from around the world who have too little growth hormone. This includes 956 patients with craniopharyngioma who were treated with growth hormone and 102 patients with craniopharyngioma who were not treated with growth hormone.

We plan to study health outcomes in these patients. We will look at how growth hormone treatment affects weight gain and development of fatty liver disease, diabetes mellitus, and heart disease in patients with craniopharyngioma. We will also look at how growth hormone treatment affects development of these same diseases in patients who do not have craniopharyngioma.

We believe this study will provide important information needed to determine whether patients with craniopharyngioma should be treated with growth hormone.

Statistical Analysis Plan:

Statistical Analysis plan:
Paired t-tests will be carried out to compare mean measures of body composition, glycemia, and liver function tests from baseline, 6 months on GH replacement, 1 year on GH replacement and annually until end of study participation. These measures include lean body mass on Dual-energy X-ray absorptiometry (DXA), body fat, Homeostatic Model Assessment of Insulin Resistance (HOMA-IR), fasting glucose, insulin, HbA1c, and BMI. If the normality assumption of the difference in the measurements is violated, the Wilcoxon signed-rank test, or the sign test in the case of a heavily skewed distribution, will be used. In addition, 95% confidence intervals for the mean difference of each marker will be reported. When comparing GH vs non-GH Craniopharyngiomas patients with respect to body composition, glycemia, and liver function test abnormalities, linear mixed effects models will be used to compare the mean change of these outcomes to take into account the repeated measure structure of the data. These models will adjust for baseline clinical and demographic factors. The Holm procedure will be used to correct for multiplicity and maintain the familywise error rate at 0.05. A similar analysis will be done for the non-CP causes of GHD cohort.
The Fisher exact test will be used to compare categorical variables and the Wilcoxon rank sum test used to compare numerical variables between two groups. Numerical results will be summarized as mean ± standard deviation and categorical results summarized as percent. Time to event estimates will be obtained by the Kaplan-Meier method, and the log-rank test used to compare the groups for mortality, when applicable. Hazard ratios will be estimated by a Cox regression model. Logistic regression will be used to analyze the association of IGF-1 SD levels and metabolic outcomes. Two-sided p values will be calculated and statistical significance set at p value <0.05. Pearson or Spearman correlation coefficients will be used to assess pairwise associations. Baseline covariates will include duration of GH replacement, duration of GHD, dose of GH, radiation therapy, sex.

Missing data plan:
Data pattern of baseline covariates with missing values will be examined using the method of [1] and in case the data is not missing completely at random, missing values will be imputed using fully conditional specification with the multivariate imputation by chained equations (MICE) algorithm under the missing at random (MAR) assumption [2,3]. Fifty or so data sets will be generated and analyzed separately, and the results combined using the formula in [4]. Similar considerations will be applied to missing data on the dependent variable in a repeated measure mixed model by investigating patterns of missingness and presence of nonignorable missing response data, see [5] for a review.
Power estimation.
We estimate power for comparing the mean liver function tests between GH and non-GH Craniopharygniomas patients. Group sample sizes of 956 GH and 102 non-GH patients achieve 80% power to detect an effect size of 0.24 in a design with 3 repeated measurements (baseline, 6 months, and 1 year) having a compound symmetry covariance structure, assuming the correlation between observations on the same subject is 0.2, and the Bonferroni correct significance level of 0.025 to correct for multiple testing of the liver function tests bilirubin, ALT, AST, and alkaline phosphatase levels (0.05 / 4 = 0.0125). If the correlation between data on the same subject is 0.6, the effect size is 0.3. Therefore, we will have enough data to be able to detect clinically meaningful differences in liver function tests.

Requested Studies:

The Global Hypopituitary Control and Complications Study
Sponsor: Eli Lilly and Company
Study ID: NCT01088399
Sponsor ID: 6448

Public Disclosures:

Odelia Cooper, Sungjin Kim. ODP643 Metabolic Effects of Growth Hormone Replacement in Craniopharyngioma and Pituitary Adenoma Patients: Analysis of the Hypopituitary Control and Complications Study (HypoCCS) Database. Journal of the Endocrine Society, Volume 6, Issue Supplement_1, November-December 2022, Page A521. doi : 10.1210/jendso/bvac150.1085