Growth Hormone Administration Improves Nonalcoholic Fatty Liver Disease in Overweight/Obesity: A Randomized Trial

Abstract

Context

Overweight and obesity are associated with relative growth hormone (GH) deficiency, which has been implicated in the development of nonalcoholic fatty liver disease (NAFLD). NAFLD is a progressive disease without effective treatments.

Objective

We hypothesized that GH administration would reduce hepatic steatosis in individuals with overweight/obesity and NAFLD.

Methods

In this 6-month randomized, double-blind, placebo-controlled trial of low-dose GH administration, 53 adults aged 18 to 65 years with BMI ≥25 kg/m² and NAFLD without diabetes were randomized to daily subcutaneous GH or placebo, targeting insulin-like growth factor 1 (IGF-1) to the upper normal quartile. The primary endpoint was intrahepatic lipid content (IHL) by proton magnetic resonance spectroscopy (1H-MRS) assessed before treatment and at 6 months.

Results

Subjects were randomly assigned to a treatment group (27 GH; 26 placebo), with 41 completers (20 GH and 21 placebo) at 6 months. Reduction in absolute % IHL by 1H-MRS was significantly greater in the GH vs placebo group (mean ± SD: −5.2 ± 10.5% vs 3.8 ± 6.9%; P = .009), resulting in a net mean treatment effect of −8.9% (95% CI, −14.5 to −3.3%). All side effects were similar between groups, except for non-clinically significant lower extremity edema, which was more frequent in the GH vs placebo group (21% vs 0%, P = .02). There were no study discontinuations due to worsening of glycemic status, and there were no significant differences in change in glycemic measures or insulin resistance between the GH and placebo groups.

Conclusion

GH administration reduces hepatic steatosis in adults with overweight/obesity and NAFLD without worsening glycemic measures. The GH/IGF-1 axis may lead to future therapeutic targets for NAFLD.

Nonalcoholic fatty liver disease (NAFLD) is a highly prevalent form of liver disease characterized by excess hepatic fat deposition and is associated with significant morbidity and mortality (1). NAFLD is an independent risk factor for cardiovascular disease and diabetes (2), and nonalcoholic steatohepatitis (NASH)-cirrhosis is expected to soon become the leading indication for liver transplantation (3). There are currently no agency-approved medications for the treatment of NAFLD.

Growth hormone (GH), which is produced by the pituitary gland and stimulates hepatic production of insulin-like growth factor-1 (IGF-1), is reduced to approximately 25% of normal in obesity (4, 5, 6). IGF-1 levels are also reduced in obesity, although to a lesser degree than GH (4, 5). GH administration in individuals with overweight/obesity is known to preferentially decrease visceral adipose tissue (VAT) and increase muscle mass (7, 8, 9). GH also acts as an anti-inflammatory cytokine (8, 9), with additional immunomodulatory properties that may be beneficial to patients with NAFLD and NASH (10).

Preclinical models, as well as data from hypopituitary patients with frank growth hormone deficiency (GHD), implicate GH and IGF-1 in the etiopathology of NAFLD and progression to NASH (11). Disruption of the GH signaling pathways in the liver of murine models leads to significant hepatic steatosis (12, 13), including in adult-onset, hepatocyte-specific GH receptor knockout mice that demonstrate a corresponding increase in hepatic de novo lipogenesis (14). IGF-1 also has been implicated in the promotion of hepatic regeneration and in reducing hepatic inflammation and fibrosis through its impact on hepatic stellate cells (15). Clinical GH replacement in patients with GHD due to hypopituitarism and concurrent NASH has been shown to improve transaminases as well as histologic steatosis and fibrosis in a small series (16). We previously demonstrated that mean peak stimulated GH was lower in individuals with overweight/obesity and NAFLD vs controls of similar age, body mass index (BMI), and sex (17).

This is the first study designed a priori to investigate the impact of growth hormone on NAFLD as a primary endpoint in otherwise healthy adults with NAFLD but without hypopituitarism. Published studies investigating the impact of GH on subjects with overweight/obesity lack gold-standard hepatic imaging such as quantification of fat by proton magnetic resonance spectroscopy (1H-MRS) (7), included subjects without NAFLD at baseline and/or only examined NAFLD-related variables as secondary endpoints (8, 9).Therefore, we sought to study the impact of low-dose daily GH administration on liver-related endpoints in individuals with overweight/obesity and NAFLD in a 6-month, randomized, double-blind, placebo-controlled trial. We hypothesized that GH would reduce hepatic steatosis by 1H-MRS as the primary study endpoint. We additionally hypothesized that GH would improve secondary hepatic endpoints, including alanine aminotransferase (ALT) and a radiographic measure of combined hepatic inflammation and fibrosis by LiverMultiScan iron-corrected T1 (cT1).

Methods

Study Design

This was a randomized, double-blind, placebo-control trial conducted at a single center (Massachusetts General Hospital, Boston, MA). This research study was approved by the Mass General Brigham Human Research Committee. Informed consent forms were signed by all subjects prior to the start of any study procedures. The trial was registered on ClinicalTrials.gov as NCT02217345.

Participants

Generally healthy adults aged 18 to 65 years with a diagnosis of NAFLD and IGF-1 levels within or below the third quartile for age were recruited between June 2, 2017, and March 11, 2021. Subjects were recruited through community advertisements, institutional research portal advertisements, and the Massachusetts General Hospital (MGH) Fatty Liver Clinic. A prior clinical diagnosis of NAFLD by imaging (ultrasound, computed tomography, or magnetic resonance imaging [MRI]) or by histology was used to confirm NAFLD for study inclusion. If there was no prior diagnosis of NAFLD, a focused ultrasound was performed at the screen visit with NAFLD diagnosis determined by an experienced radiologist. Exclusion criteria included any history of diabetes mellitus (except for a history of gestational diabetes), use of diabetes medications, aminotransferase levels >10× normal, creatinine >2× normal, prior or current malignancy, use of oral estrogen products, active carpal tunnel syndrome, cirrhosis or other forms of chronic liver disease, excessive alcohol ingestion (>3 drinks/day in men and >2 drinks/day in women), contraindication to MRI imaging, pregnancy or breastfeeding, recent weight loss surgery or weight instability, use of weight loss medications in the past 6 months, known hypothalamic or pituitary disease affecting the GH axis, or regular use of drugs within the prior year that are known to cause hepatic steatosis (including oral steroids, methotrexate and tamoxifen). After written informed consent, 131 individuals were screened for eligibility with 53 randomized and 41 who completed the study (Fig. 1, Consort Diagram).

Figure 1.

Consort diagram.

Randomization and Masking

Subjects were assigned 1:1 to receive daily subcutaneous GH (Genotropin, Pfizer Inc.) or identical placebo. Randomization was performed by the MGH Research Pharmacy via computer generated list in blocks of 2 and stratified by sex. The GH and placebo vials were identical to ensure double blinding for investigators and participants. An unblinded study monitor who was not involved in any other aspect of the trial received IGF-1 level results directly and made true (GH) or sham (placebo) dose adjustments to maintain double blinding.

Procedures

Fasting screening laboratory tests included hemoglobin A1c (HbA1c), glucose, basic metabolic panel, liver function panel [including alanine aminotransferase (ALT)], complete blood count, thyroid-stimulating hormone, and IGF-1 levels (liquid chromatography–tandem mass spectrometry [LC-MS/MS], Quest Diagnostics Nichols Institute, San Juan, Capistrano, CA). The baseline visit was conducted after an 8-hour overnight fast. Blood samples were drawn for glucose, insulin, high-sensitivity C-reactive protein (hsCRP) (immunoturbidimetric assay, Roche Diagnostics, Indianapolis, IN), and lipid panel. Intrahepatic lipid content (IHL) was assessed by single voxel breath-hold 1H-MRS in the right hepatic lobe using a 3.0-Tesla MR scanner as previously described (Vida, Siemens Medical Systems, Erlangen, Germany; Discovery, General Electric, Waukesha, WI, USA) with a coefficient of variation (CV) for same-day measurements at our institution of 8% for IHL quantification (18). 1H-MRS data fitting was performed using LCModel (version 6.1-4A, S. Provencher, Oakville, Ontario, CA) (19). IHL estimates were automatically scaled to unsuppressed water peak (4.7 ppm) and are expressed as percent IHL. LiverMultiScan multiparametric MRI data were acquired in the same session as the 1H-MRS. Expiratory breath-hold (<10 seconds) and electrocardiography/pulse gating were utilized to minimize motion from respiration or pulsation. T1 mapping for inflammation and fibrosis was performed as previously published (20). T2* mapping for hepatic iron quantification was also performed. T1 mapping for assessment of liver fibrosis and inflammation was corrected for hemosiderosis as measured from T2* mapping, yielding an iron-corrected T1 (cT1) value. Hepatic fat was also quantified by proton density fat fraction (PDFF) (20). All postprocessing of the obtained images was performed by trained analysts who were blinded to clinical data and randomization assignment. Body composition, including fat and lean body mass, were measured by dual-energy x-ray absorptiometry (DXA) (Hologic Discovery A or equivalent, Hologic, Marlborough, MA, USA) with a precision of <2% for lean mass (21). A GH-releasing hormone (GHRH)–arginine stimulation test was performed per standard protocol as previously published (17): GHRH (1 μg/kg) plus arginine (0.5 g/kg; max 30 grams) was administered intravenously, and GH levels were drawn at 0, 30, 60, 90, and 120 minutes. GHRH was obtained from Stratum Medical Corporation (San Diego, CA) and administered under a research IND. Serum GH was measured by immunoassay (Quest Diagnostics, Marlborough, MA).

After baseline testing, subjects were instructed on proper subcutaneous injection technique with the injection pen device, and the first injection was observed by the study team. Starting doses were 0.2 mg/day in men and 0.3 mg/day in women. Dose titrations were performed based on IGF-1 levels by the unblinded study monitor at the 1-month, 2-month, and 3-month visits. All members of the study team except the unblinded monitor were blinded to follow-up IGF-1 levels. The GH dose was increased by 0.1 to 0.2 mg/day with a goal IGF-1 in the upper quartile of normal for age with dose decreases by 0.1 to 0.2 mg/day in the case of an elevated IGF-1 or GH-related side effects. Sham dose adjustments were made in the placebo subjects to maintain the double blind. All subjects were asked to complete daily drug administration diaries.

Fasting glucose, aminotransferases, and IGF-1 levels were measured as safety labs at the 1-month, 2-month, 3-month, and 6-month visits. HbA1c was repeated at the 3-month and 6-month visits. Hepatic MRI, DXA, hsCRP, and insulin assessments were repeated at the 6-month visit. Injections were discontinued after the final study visit.

Outcomes

The primary endpoint was absolute change in % IHL content by 1H-MRS between the GH and placebo groups from baseline to 6 months. Secondary endpoints included baseline to 6-month change in IHL content by PDFF, combined radiographic inflammation and fibrosis (cT1), ALT, body composition, and hsCRP between groups. Prior to unblinding, all subjects with ≥ 3% weight loss (GH, n = 3; placebo, n = 5) were identified, as this degree of weight loss has been shown to independently improve NAFLD endpoints (1). These subjects were then excluded for a secondary analysis of weight-stable individuals in order to isolate the effects of GH on study endpoints. Adverse events, including all potential side effects of GH treatment, were assessed by study investigators at all study visits and at a follow-up phone call 1 month after drug discontinuation. Glycemic measures were monitored throughout the study as described above with prespecified discontinuation criteria of fasting glucose ≥ 126 mg/dL and/or HbA1c ≥ 6.5%.

Statistical Analysis

JMP Statistical Database Software (version 16; SAS Institute, Cary, NC) was used for statistical analyses. Variables with a non-normal distribution were log-transformed prior to analysis. Analysis of variance (ANOVA) was used to compare baseline data between the 2 groups. The primary analysis was a pooled random slopes model analysis of treatment effect across sexes, with the weights equaling the frequency of men and women in the sample with the primary endpoint of change in hepatic steatosis by 1H-MRS. Mixed-model analysis was utilized to eliminate complete-case bias. For all analyses, an intent-to-treat analysis was performed. Guidelines developed by the National Research Council regarding missing data were applied. The random slopes model (primary analysis) assumed that data were missing at random. Our study was powered for n = 50 randomized and a 20% dropout rate to 40 completers, for which we had 90% power to detect a 6.5% change in IHL at a 2-tailed P value of .05. This was based on the assumption that the SD of the response variable is 6.1% as observed in prior studies (22). Secondary endpoints included the change in hepatic inflammation and fibrosis as measured by cT1, ALT, hsCRP, and body composition. Multivariable least squares and stepwise regression analysis were used to control for potential confounders, including sex, age, VAT by DXA, and homeostatic model assessment for insulin resistance (HOMA-IR) and to model predictors of endpoints. Statistical significance was defined as a 2-tailed P ≤ 0.05. Data are reported as mean ± SD or n (%), unless otherwise noted. Significant differences in rates of adverse events between drug-exposed individuals in the GH vs placebo groups were calculated using a Fisher exact test. Study progress and safety data were reviewed by a Data Safety and Monitoring Board every 4 to 6 months.

Results

Between June 2, 2017, and March 11, 2021, 131 people were screened for study eligibility and 53 were randomly assigned to a treatment group (27 GH and 26 placebo). One subject assigned to the GH group did not complete any baseline procedures and thus did not contribute data to the baseline cohort. Baseline characteristics of the GH (n = 26) and placebo groups (n = 26) are reported in Table 1. The baseline cohort (n = 52) was 50% female with a mean (SD) age of 46.0 ± 12.1 years and mean BMI of 33.0 ± 5.7 kg/m². Additional clinical characteristics of the study population are reported in Supplementary Table S1 (23). There were 41 completers (20 GH and 21 placebo) at 6 months who had repeat assessment of the primary endpoints.

Table 1.

Baseline characteristics

	Whole cohort (n = 52) (mean ± SD)	GH (n = 26) (mean ± SD)	Placebo (n = 26) (mean ± SD)
Demographics
Age (years)	46.0 ± 12.1	46.1 ± 11.8	45.8 ± 12.5
Female sex (% female)	26 (50%)	13 (50%)	13 (50%)
BMI (kg/m2)	33.0 ± 5.7	33.6 ± 5.1	32.3 ± 6.3
Weight (kg)	95.9 ± 20.7	99.2 ± 20.1	92.7 ± 21.2
Iliac waist circumference (cm)	108 ± 13	110 ± 11	107 ± 14
Race
Asian (n, %)	7 (13%)	5 (19%)	2 (8%)
Black (n, %)	2 (4%)	1 (4%)	1 (4%)
White (n, %)	39 (75%)	19 (73%)	20 (77%)
More than one race (n, %)	3 (6%)	1 (4%)	2 (8%)
Unknown (n, %)	1 (2%)	0	1 (3%)
Ethnicity
Hispanic (n, %)	2 (4%)	0 (0%)	2 (8%)
Hepatic endpoints
IHL (%)	21.4 ± 14.5	24.7 ± 14.8	18.3 ± 13.8
IHL by PDFF (%)	12.2 ± 8.9	13.5 ± 8	10.9 ± 9.7
cT1 (ms)	848.9 ± 90.7	860.3 ± 87.4	837.5 ± 94.4
Body composition (DXA)
Total body fat (%)	37.2 ± 6.8	37.6 ± 7.2	36.9 ± 6.5
Total lean body mass (kg)	58.7 ± 13.3	60.7 ± 14.1	56.8 ± 12.4
Estimated VAT Area	148.0 ± 45.6	152.0 ± 45.1	144.0 ± 46.7
Laboratory values
ALT (IU/L)	37 ± 42	34 ± 17	40 ± 57
AST (IU/L)	25 ± 12	24 ± 8	25 ± 16
Peak stimulated GH (ng/mL)	9.2 ± 6.1	7.9 ± 4.2	10.4 ± 7.3
IGF-1 (ng/mL)	148 ± 53	147 ± 48	148 ± 58
IGF-1 Z-score	− 0.1 ± 0.7	0.0 ± 0.6	−0.1 ± 0.7
HbA1c (%)	5.4 ± 0.3	5.5 ± 0.3	5.4 ± 0.2
Fasting glucose (mg/dL)	84 ± 8	86 ± 8	83 ± 8
Insulin (mIU/L)	9.5 ± 5.1	10.6 ± 5.1	8.4 ± 5.0
HOMA-IR	2.0 ± 1.1	2.3 ± 1.1	1.8 ± 1.1
Lipids
Cholesterol total (mg/dL)	186 ± 26	184 ± 26	189 ± 26
Triglycerides (mg/dL)	130 ± 57	131 ± 70	129 ± 42
HDL cholesterol (mg/dL)	46 ± 11	46 ± 11	46 ± 10
VLDL cholesterol calc (mg/dL)	26 ± 12	27 ± 15	26 ± 8
LDL cholesterol calc (mg/dL)	116 ± 24	113 ± 23	119 ± 25

Data are reported as mean ± SD.

Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; DXA, dual-energy x-ray absorptiometry; GH, growth hormone; HOMA-IR, homeostatic model assessment of insulin resistance; IHL, intrahepatic lipid content; PDFF, proton density fat fraction; VAT, visceral adipose tissue.

There were significant inverse correlations at baseline (pretreatment) between peak stimulated GH and IHL by 1H-MRS (R = −0.29, P = .04), IHL by PDFF (R = −0.33, P = .02), and cT1 (R = −0.37, P = .01) but not ALT (R = − 0.09, P = .5). After adjusting for age, sex, and BMI, peak stimulated GH was still significantly associated with IHL by 1H-MRS (β = −.42, P = .02), IHL by PDFF (β = −.44, P = .01), and cT1 (β = −.52, P = .004). IGF-1 levels at baseline did not correlate with any of these variables.

Mean IGF-1 levels and IGF-1 Z-scores significantly increased in the GH vs placebo groups over 6 months of drug administration (delta mean IGF-1 106.1 ± 52.5 ng/mL vs −11.8 ± 33.0 ng/mL, P = .007 and delta IGF-1 Z-score 1.3 ± 0.6 vs −0.1 ± 0.5, P < .0001). The resulting mean IGF-1 and IGF-1 Z-scores were 256.2 ± 59.3 ng/mL and 1.4 ± 0.6 in the GH group (Table 2, Fig. 2). The mean 6-month GH dose in the active group was 0.5 ± 0.2 mg daily (range, 0.1-0.9 mg daily), with a mean dose of 0.5 ± 0.2 mg/day in men vs 0.6 ± 0.2 mg/day in women (P = .1). The mean (± SD) total missing doses in the GH group was 11 ± 10 doses (range, 0-36 doses) out of a potential approximately 180 doses during the 6-month study. There was no difference between the mean number of missed doses in the GH vs placebo groups (P = .9).

Table 2.

Effects of GH on hepatic and metabolic endpoints at 6 months

	6-month change
	GH (n = 20) mean ± SD	Placebo (n = 21) mean ± SD	P value GH vs placebo
BMI and body composition
BMI (kg/m2)a	−0.2 ± 1.2	−0.1 ± 1.2	.7
Weight (kg)	−0.6 ± 3.6	−0.2 ± 3.5	1.0
Iliac waist circumference (cm)a	−0.8 ± 2.9	0.8 ± 4.9	.6
DXA estimated VAT area (cm2)	−9.9 ± 9.3	−0.1 ± 19.8	.05
DXA total fat (%)	−0.7 ± 1.3	−0.2 ± 2.2	.3
DXA total lean body Mass (kg)	0.8 ± 2.3	0.5 ± 2.3	.7
Android/Gynoid ratio	−0.02 ± 0.05	−0.02 ± 0.05	.7
Hepatic endpoints
IHL by MRS (absolute %)a	−5.2 ± 10.5	3.8 ± 6.9	.009c
IHL by MRS (relative %)b	−22.1 ± 40.4	42.1 ± 87.1	.004c
IHL by PDFF (absolute %)	−3.4 ± 3.6	−0.4 ± 3.3	.05c
cT1 (ms)a	−7.2 ± 58.1	19.1 ± 47.8	.2c
ALT (IU/L)a	−10 ± 13	−2 ± 12	.002c
Laboratory values
IGF-1 (ng/mL)	106 ± 53	−12 ± 33	.007
IGF-1 Z-Score	1.3 ± 0.6	−0.1 ± 0.5	<.0001
HbA1c (%)	0.0 ± 0.2	0.0 ± 0.2	.8
Fasting glucose (mg/dL)	−1 ± 8	1 ± 7	.4
Insulina	1.1 ± 6.3	0.2 ± 4.0	1.0
HOMA-IRa	0.2 ± 1.4	0.0 ± 0.9	.9
hsCRP (mg/L)a	−0.8 ± 0.9	0.3 ± 1.7	.003
Lipids
Cholesterol total (mg/dL)	10 ± 20	17 ± 30	.4
Triglycerides (mg/dL)a	−2 ± 45	9 ± 52	.9
HDL (mg/dL)a	3 ± 7	1 ± 4	.3
VLDL (mg/dL)	−1 ± 10	1 ± 10	.4
LDL (mg/dL)	7 ± 19	15 ± 26	.3

Data reported as mean ± SD.

Abbreviations: ALT, alanine aminotransferase; BMI, body mass index; DXA, dual-energy x-ray absorptiometry; GH, growth hormone; 1H-MRS, proton magnetic resonance spectroscopy; HbA1c, glycated hemoglobin; HOMA-IR, homeostatic model assessment of insulin resistance; IHL, intrahepatic lipids; PDFF, proton density fat fraction; PDFF, proton density fat fraction.

^a Indicates non-normally distributed variable where P value was derived from mixed-model analysis after log transformation.

^b Indicates non-normally distributed variable where P value was derived from t test after log transformation.

^c Indicates significant difference (P < .05) in GH vs placebo groups in sensitivity analysis of hepatic endpoints excluding individuals with >3% weight loss (n = 3 GH, n = 5 placebo).

Figure 2.

Change in mean IGF-1 Z-Scores in GH vs placebo treated subjects over 6 months. Data are reported as mean ± SEM.

As expected, there was no difference in change in weight or BMI between the GH and placebo groups during the 6-month treatment period (mean change in weight −0.6 ± 3.6 kg vs −0.2 ± 3.5 kg, P = 1.0 and mean change in BMI −0.2 ± 1.2 kg/m² vs −0.1 ± 1.2 kg/m², P = .7 for GH and placebo groups, respectively) (Table 2). Eight subjects identified prior to unblinding experienced ≥3% weight loss (3 GH and 5 placebo, P = .7).

The primary endpoint, IHL by 1H-MRS, improved in the GH vs placebo group (absolute change −5.2 ± 10.5% vs 3.8 ± 6.9%, P = .009), resulting in a net treatment effect of −8.9% (95% CI, −14.6 to −3.3%; Fig. 3A) (Table 2). The relative reduction of IHL from baseline by 1H-MRS was −22.1 ± 40.4% in the GH group and 42.1 ± 87.1% in the placebo group, P = .004). Effects on secondary endpoints were as follows: hepatic steatosis by PDFF was significantly reduced in the GH vs placebo group (absolute change −3.4 ± 3.6% vs −0.4 ± 3.3%, P = .05) and serum ALT decreased in the GH vs placebo group (−10.3 ± 12.6 IU/L vs −1.6 ± 11.9 IU/L, P = .002; Fig. 3B). Mean change in cT1 on univariate analysis was similar between the GH and placebo group (−7.2 ± 58.1 ms vs 19.1 ± 47.8 ms, P = .2; Fig. 3C).

Figure 3.

Change in intrahepatic lipid content (IHL) by 1H-MRS (A), ALT (B), and cT1 (C) in response to 6 months of GH vs placebo. Data reported as mean ± SEM. P values reported for IHL and ALT are after log transformation.

The effect of GH vs placebo on primary and secondary endpoints remained significant in multivariable models adjusting for age, sex, and change in BMI, as follows: IHL by 1H-MRS (P = .001, accounting for 23% of variability by eta squared) and ALT (P = .04, accounting for 10% of variability by eta squared). There was also a trend toward a treatment effect of GH vs placebo on cT1 in this model (P = .07, accounting for 5% of the variability by eta squared).

A secondary analysis was performed to isolate the effects of GH on NAFLD endpoints independent of weight loss experienced by subjects in both groups during the trial. Subjects with ≥ 3% weight loss, which has been shown to independently improve NAFLD endpoints (1, 24), were identified prior to unblinding and were excluded in this secondary analysis (n = 8, 3 GH and 5 placebo). There were significant improvements in IHL by 1H-MRS (absolute change −1.9 ± 6.2% vs 5.8 ± 6.1% P = .01), IHL by PDFF (absolute change −2.5 ± 2.7% vs 0.6 ± 1.5%, P = .003), ALT (−10.2 ± 13.4 IU/L vs −0.4 ± 13.1 IU/L, P = .001), and cT1 (0.1 ± 53.2 ms vs 35.7 ± 32.8 ms, P = .04) in the GH vs placebo group.

There was significant improvement in other secondary endpoints in the GH vs placebo group, including a significant reduction in VAT area (−9.9 ± 9.3 cm² vs −0.1 ± 19.8 cm², P = .05) and mean hsCRP (−0.8 ± 0.9 mg/dL vs 0.3 ± 1.7 mg/dL, P = .003). There were no significant differences in change in BMI, lean body mass, lipid levels, glycemic measures or insulin resistance (HbA1c, fasting glucose, and HOMA-IR) between the GH and placebo groups over the 6-month trial.

Absolute change in IGF-1 was significantly associated with improvement in IHL by 1H-MRS (R = −0.45, P = .005) when examined across the whole cohort, which remained significant when adjusted for age, sex, and change in BMI. There was no association between change in IGF-1 level and improvement in ALT (R = −0.20, P = .2) or cT1 (R = −0.25, P = .1).

There were no treatment-related serious adverse events or safety concerns identified during the study period. There was a higher rate of non-clinically significant lower extremity edema in the GH vs placebo group (21% vs 0%, P = .02) that did not lead to study drug discontinuation in any participant. There were otherwise no differences between groups in injection site discomfort or bruising, carpal tunnel syndrome, myalgias, numbness or tingling, joint pain/stiffness, or headaches (Table 3). No participants dropped out of the study due to glycemic-related issues, and no subjects developed a clinical diagnosis of diabetes, which was a prespecified stop criterion for the study (HbA1c ≥ 6.5% or fasting glucose of ≥ 126 mg/dL). There was no evidence of significant worsening of glycemic status with GH over the course of the 6-month treatment period (Supplementary Table S2) (23). Notably, no study completers were treated with antihyperglycemic medications during the study period.

Table 3.

Adverse events

	Growth hormone (n = 24)	Placebo (n = 25)	P value
Edema	5 (21%)	0 (0%)	.02
Injection site discomfort or bruising	4 (17%)	5 (20%)	1.0
Joint pain or stiffness	4 (17%)	6 (24%)	.73
Myalgias	3 (13%)	2 (8%)	.67
Numbness or tingling	3 (13%)	2 (8%)	.67
Headache	3 (13%)	4 (16%)	1.0
Allergic reaction	1 (4%)	0 (0%)	.49
Hyperglycemia	0 (0%)	0 (0%)	1.0
Carpal tunnel syndrome	0 (0%)	1 (4%)	1.0

Adverse events are presented as a percentage of subjects in each group who were exposed to drug or placebo.

There were 4 dropouts due to adverse events or symptoms. One subject in the GH group was discontinued immediately after an allergic reaction to the study medication (self-resolving, no emergency department visit required). A second subject in the GH group was discontinued just after the 3-month visit due to an unrelated serious adverse event (admission for orthopedic surgery after trauma). Two subjects were discontinued from the placebo group due to adverse events, including one who chose to discontinue due to concern that the study drug could be the cause of her weight gain (n = 1) and one with diffuse joint pain requiring rheumatologic evaluation (n = 1).

Discussion

Data from this randomized, double-blind, placebo-controlled study demonstrate that daily, subcutaneous GH administration in subjects with overweight/obesity and NAFLD led to a 9% absolute reduction in liver fat compared to placebo. There was additionally a 22% relative reduction in liver fat by gold-standard 1H-MRS and a reduction in hepatocellular damage as assessed by ALT. These findings were robust, as improvements in liver fat were demonstrated when measured by both 1H-MRS and PDFF and persisted when adjusting for age, sex, and change in BMI and in a sensitivity analysis excluding all subjects with ≥3% weight loss, which itself has been shown to result in significant improvements in hepatic steatosis (24). There was additionally a positive effect of GH when compared to placebo on radiographic inflammation and fibrosis (cT1) in the sensitivity analysis designed to isolate the effects of GH from those of weight loss on hepatic endpoints. The mean GH dose in this study was similar to (25) or lower than (8, 9) prior studies of GH administration in subjects with overweight/obesity who were not required to have NAFLD to participate. This study did not identify any safety concerns regarding GH administration in subjects specifically selected for overweight/obesity and NAFLD without diabetes mellitus.

GH has known lipolytic and anti-inflammatory effects (8, 9), and evidence in patients with hypopituitarism and severe GHD implicates the GH/IGF-1 axis in the pathophysiology of NAFLD and progression to NASH. A study of 66 adults with hypopituitarism with GHD were shown to have a higher rate of NAFLD and higher ALT levels than age-, sex-, and BMI-matched peers (16). Additionally, 5 patients with hypopituitarism, GHD, and NASH demonstrated improvements in histologic steatosis, histologic fibrosis, and serum markers of inflammation and fibrosis after 6 to 12 months of physiologic GH replacement (16). While a subsequent randomized trial of physiologic GH replacement vs no treatment in adults with hypopituitarism and GHD did not show improvement in IHL with treatment, the population studied was unselected for NAFLD and less than one-quarter of subjects had NAFLD at baseline, leaving little room for improvement in intrahepatic lipid content (26).

Obesity is an established model of relative GHD in which GH secretion decreases linearly with weight (4, 5, 6), and studies have demonstrated reversal of GH secretory abnormalities with weight loss (27). We previously demonstrated that lower peak stimulated GH levels in adults with overweight/obesity were associated with higher levels of liver fat assessed by 1H-MRS and greater hepatocellular damage assessed by ALT (17). We also demonstrated that lower serum IGF-1 levels were associated with histologic NASH, inflammation, and fibrosis (28). Therefore, in contrast to classic pharmacologic approaches, this study targets normalization of an endogenous hormone deficiency by investigating GH administration in subjects with relative GH deficiency of obesity and NAFLD.

One open-label study of GH administration vs no drug over 6 months in adolescents with obesity without pituitary disorders did not demonstrate a significant change in intrahepatic lipid content, although more subjects in the GH group had resolution of NAFLD (defined as <5% hepatic steatosis by 1H-MRS) compared to the untreated group (29). The GHRH analogue tesamorelin, which augments GH in a pulsatile fashion, has been shown to improve hepatic steatosis in individuals with HIV-associated NAFLD (30) but has not been studied in otherwise healthy individuals with overweight/obesity and NAFLD. Therefore, this is the first randomized, double-blind, placebo-controlled investigation of the effects of daily subcutaneous GH in otherwise healthy adults with NAFLD, and it demonstrated that GH augmentation decreases hepatic steatosis and ALT and improves hsCRP and visceral adiposity in otherwise healthy adults with overweight/obesity and NAFLD.

While GH is known to decrease visceral fat and increase muscle mass, physiologic GH administration is not expected to induce weight loss (7). However, as little as 3% weight loss has been shown to improve hepatic steatosis, and higher degrees of weight loss (7%-10%) lead to improvement in NASH and fibrosis (1). Although change in weight was not different between the GH and placebo groups over 6 months, we performed a secondary analysis excluding individuals with more than 3% weight loss to isolate the effects of GH on NAFLD endpoints. In this secondary analysis that isolated the effects of GH from weight loss, the effect of GH on improvement in hepatic steatosis and ALT remained significant and the effect of GH on improvement in inflammation and fibrosis by cT1 became significant.

Despite its overall weight neutrality, GH had a significant impact on reduction in hepatic steatosis that is comparable to weight loss cohorts in other trials. For example, a lifestyle intervention compared to a control group in adults with overweight/obesity demonstrated a treatment effect of a 4.5% absolute reduction in intrahepatic lipids by MRS and a 14-point reduction in ALT in the context of a net weight reduction of 5 kg (−5.6 kg in intervention group vs −0.6 kg in control group) (22). In a trial of the impact of liraglutide for NAFLD, those in the middle tertile of weight loss (mean loss of 3 kg) had an absolute reduction of approximately 4.8% of intrahepatic lipid content by MRS while the highest tertile of weight loss (mean loss of 9.6 kg) had a 10.1% absolute reduction in intrahepatic lipids (31). In this context, daily physiologic GH replacement in the current trial had a treatment effect of an 8.9% reduction in intrahepatic lipid content, demonstrating a significant direct effect on reduction in liver fat in the absence of weight loss. A 22% relative reduction in liver fat in the GH group is also notable given that a 30% relative reduction in liver fat by PDFF has been associated with regression of hepatic fibrosis in published studies (32). Therefore, identification of the molecular mechanisms of the impact of GH on these NAFLD endpoints may lead to therapies that could be complementary to weight loss recommendations and/or appropriate for individuals with lean NAFLD.

Strengths of this study included the randomized, double-blind, and placebo-controlled design, requirement of NAFLD diagnosis at study entry, and the use of gold-standard assessment of hepatic steatosis (1, 33). We did not include subjects with diabetes mellitus in this study in order to select for a more uniform population, avoid confounding by treatment with diabetes-related medications, and to avoid potential adverse hyperglycemic effects of GH in a more susceptible population. We are therefore limited in our ability to extrapolate efficacy and safety to patients with diabetes mellitus. However, we did include subjects with prediabetes, and despite the theoretical concern for GH to cause hyperglycemia, no subjects were discontinued from this trial due to hyperglycemia or new diagnosis of diabetes mellitus and there were no significant effects seen on any glucose or insulin-related endpoints. Exclusion of subjects with type 2 diabetes may have also selected for a cohort that was not enriched for a higher severity of hepatic inflammation and fibrosis at baseline. Insulin resistance was assessed by HOMA-IR, which is an imperfect measure. Finally, we are lacking histologic assessment pre- and posttreatment in this cohort due to funding limitations.

In conclusion, low-dose GH administration with augmentation to the upper normal range in otherwise healthy adults with overweight/obesity and NAFLD improves hepatic steatosis and hepatocellular damage. Low-dose GH administration in this population with overweight/obesity and NAFLD did not cause adverse glycemic effects or an increase in insulin resistance. Future research should focus on identifying the mechanisms of the effects of GH and IGF-1, which likely have unique and independent mechanisms to reduce hepatic steatosis and ameliorate hepatic inflammation and fibrosis. In addition to elucidating underlying pathophysiology of this disease process, identification of the molecular mechanisms of the GH/IGF-1 axis in NAFLD/NASH has the potential to lead to the development of novel therapeutics. This area of investigation is highly relevant at this time given the high prevalence of NAFLD/NASH in obesity, serious consequences of NASH cirrhosis, and lack of effective treatments for NAFLD.

Abbreviations

1H-MRS

proton magnetic resonance spectroscopy

ALT

alanine aminotransferase

cT1

iron-corrected T1

DXA

dual-energy x-ray absorptiometry

GH

growth hormone

GHD

growth hormone deficiency

GHRH

growth hormone-releasing hormone

HbA1c

glycated hemoglobin

HOMA-IR

homeostatic model assessment for insulin resistance

hsCRP

high-sensitivity C-reactive protein

IGF-1

insulin-like growth factor 1

IHL

intrahepatic lipids

NAFLD

nonalcoholic fatty liver disease

NASH

nonalcoholic steatohepatitis

PDFF

proton density fat fraction

VAT

visceral adipose tissue

Funding

The work was conducted with support from the following grants: National Institutes of Health (NIH) K23 DK113220 (Dichtel), NIH K24 HL092902 (Miller), NIH K24 DK109940 (Bredella), Massachusetts General Hospital Claflin Distinguished Scholar Award (Dichtel), Harvard Medical School Eleanor and Miles Shore Award (Dichtel), NIH 8 UL1 TR000170 (Harvard Clinical and Translational Science Center, from the National Center for Advancing Translational Sciences), NIH 1UL1TR0011-02 and NIH 1UL1TR002541-01. The content is solely the responsibility of the authors and does not necessarily represent the official views of Harvard Catalyst, Harvard University and its affiliated academic healthcare centers, or the National Institutes of Health. Additional grants to authors during this study period include: K23 DK115903 (Haines), T32 DK007028 (Kimball/Colling), K23 DK113252 (Long), the Doris Duke Charitable Foundation Grant #2019085 (Long).

Disclosures

Pfizer donated identical growth hormone and placebo (L.E.D./K.K.M) and Perspectum Ltd. donated LiverMultiScan analysis for this trial (L.E.D.). L.E.D. has also received research support from Perspectum Ltd. and Lumos Pharma per investigator-initiated requests, has research support from Recordati, and has equity in Marea Therapeutics. L.E.D. is a Mass General Brigham Innovation Fellow hosted by Third Rock Ventures, a venture capital firm. She remains full-time at MGH during the period of this educational program (anticipated October 1, 2022, to September 30, 2024). L.E.D.'s financial interests were reviewed and are managed by MGH and Mass General Brigham in accordance with their conflict-of-interest policies. K.K.M has received study medication and investigator-initiated research grants from Amgen and has equity in Bristol-Myers Squibb (BMS), General Electric, Boston Scientific, and Becton Dickinson. K.E.C. serves as a consultant or advisory board member for BMS, Novo Nordisk, Gilead, and Theratechnologies and has received grant support from Novartis, BMS, Boehringer-Ingelheim. T.G.S. has received grants to MGH from Amgen for work unrelated to the current study. After the completion of this trial, M.T.L. began working at Novo Nordisk (Obesity, NASH, & Devices, Medical & Science, Clinical Drug Development, Novo Nordisk, Søborg, Denmark) as of July 2022. The remaining authors have no disclosures to report.

Data Availability

Some or all datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

Clinical Trials Information

ClinicalTrials.gov identification no. NCT02217345.