Effects of Testosterone Supplementation on Ghrelin and Appetite During and After Severe Energy Deficit in Healthy Men

Abstract

Background

Severe energy deficits cause interrelated reductions in testosterone and fat free mass. Testosterone supplementation may mitigate those decrements, but could also reduce circulating concentrations of the orexigenic hormone ghrelin, thereby exacerbating energy deficit by suppressing appetite.

Objective

To determine whether testosterone supplementation during severe energy deficit influences fasting and postprandial ghrelin concentrations and appetite.

Design and methods

Secondary analysis of a randomized, double-blind trial that determined the effects of testosterone supplementation on body composition changes during and following severe energy deficit in nonobese, eugonadal men. Phase 1 (PRE-ED): 14-day run-in; phase 2: 28 days, 55% energy deficit with 200 mg testosterone enanthate weekly (TEST; n = 24) or placebo (PLA; n = 26); phase 3: free-living until body mass recovered (end-of-study; EOS). Fasting and postprandial acyl ghrelin and des-acyl ghrelin concentrations and appetite were secondary outcomes measured during the final week of each phase.

Results

Fasting acyl ghrelin concentrations, and postprandial acyl and des-acyl ghrelin concentrations increased in PLA during energy deficit then returned to PRE-ED values by EOS, but did not change in TEST (phase-by-group, P < 0.05). Correlations between changes in free testosterone and changes in fasting acyl ghrelin concentrations during energy deficit (ρ = -0.42, P = 0.003) and body mass recovery (ρ = -0.38; P = 0.01) were not mediated by changes in body mass or body composition. Transient increases in appetite during energy deficit were not affected by testosterone treatment.

Conclusions

Testosterone supplementation during short-term, severe energy deficit in healthy men prevents deficit-induced increases in circulating ghrelin without blunting concomitant increases in appetite.

Clinical Trials Registration

www.clinicaltrials.gov NCT02734238 (registered 12 April 2016).

hypothalamic-pituitary-gonadal axis, hypogonadism, androgen, military, weight cycling

Severe energy deficits (~50%-100% of total daily energy expenditure) are common in military personnel during training [1, 2]. These deficits result from an inability or unwillingness to increase energy intake to match energy expenditures ranging from 4000 to 7000 kcal/d [3–5]. Consequences include total body mass (TBM) and fat free mass losses [6–10], and decreased physical performance [1]. Substantial reductions in circulating testosterone likely contribute to those decrements [6, 11–14]. Although body mass is regained and testosterone concentrations are restored by refeeding [6, 15], the amount of body mass regained has been reported to exceed the amount of TBM lost and be disproportionately comprised of fat mass (FM) [6, 16–18]. This “weight and fat overshoot” is thought to be caused, in part, by a physiologic drive to fully recover body protein stores [18–20], which may suggest reduced testosterone is a contributing factor.

Energy deficits also elicit adaptive responses in peripheral and central regulators of energy balance that stimulate appetite and promote TBM regain [21, 22]. One potential endocrine mediator of this response is ghrelin [23], a peptide hormone secreted primarily from the gastric mucosa in an acylated form that is rapidly converted into an unacylated (des-acyl) form that composes the majority of ghrelin in circulation [24, 25]. Acyl ghrelin exerts multiple physiologic effects [26], which include modulating glucose homeostasis and stimulating appetite and adipogenesis [27–31]. Des-acyl ghrelin also appears to regulate these outcomes, although effects may either reinforce or antagonize those of acyl ghrelin depending on the experimental context [32]. Nonetheless, circulating total (acyl + des-acyl) ghrelin increases in response to energy deficit [23, 33], and may contribute to a drive to eat that promotes regain of TBM and FM [21].

Circulating ghrelin concentrations may also be regulated by testosterone. Ghrelin is expressed in human testes [34, 35], and testosterone supplementation suppresses circulating total ghrelin in prepubertal boys and weight-stable, nonobese, eugonadal men [36, 37]. This effect, if present in nonobese men during severe energy deficit, may have the unintended and undesirable consequence of blunting energy deficit-induced increases in ghrelin and, subsequently, appetite. This could, in turn, exacerbate energy deficit. Recovery of TBM and FM during refeeding might also be slowed. Although effects of testosterone and other androgens on appetite regulation are poorly characterized [38], those responses would be consistent with studies reporting that testosterone supplementation is associated with gradual reductions in FM [39], and in some cases TBM [40]. However, testosterone supplementation has also been reported to increase circulating total ghrelin during energy balance [41] and have no effect during energy deficit [42] in men with overweight or obesity and low testosterone. The effect of testosterone supplementation on circulating ghrelin therefore remains uncertain and may depend on population demographics and/or be influenced by energy balance. Whether testosterone supplementation affects acyl and des-acyl ghrelin concentrations differently, and if any effects influence appetite and eating behavior has not been examined.

Our group recently published the results of a study designed to determine the effects of testosterone supplementation on changes in body mass and body composition during and following 28 days of severe energy deficit in healthy, nonobese, eugonadal young men [43]. Study results demonstrated that supplemental testosterone increased lean body mass during energy deficit. Herein, we report secondary outcomes included in the design of that study that addressed tertiary study objectives. Those objectives were to determine the effects of testosterone supplementation on circulating acyl and des-acyl ghrelin concentrations and appetite during and after severe energy deficit, and to explore relationships between changes in testosterone, ghrelin, appetite, TBM, and body composition during severe energy deficit and TBM recovery [44].

1. Materials and Methods

This study addressed tertiary objectives included in the Optimizing Performance for Soldiers study [43–45]. All outcomes presented herein were defined in the original study protocol. The study was approved by the Pennington Biomedical Research Center (PBRC) institutional review board and the Human Research Protection Office of the US Army Medical Research and Materiel Command. Participants provided written informed consent, and the study was registered at www.clinicaltrials.gov as NCT02734238.

A. Subjects

Healthy, physically active men aged 18–39 years (mean ± SD = 25 ± 1 year), with total testosterone concentrations within normal physiological range (300 – 1000 ng/dL), and who met age-specific US Army body composition standards were recruited. Full inclusion and exclusion criteria and details of the recruitment process are published elsewhere [44].

B. Study Design

The study used a randomized, double-blind, placebo-controlled design. Full details of the Optimizing Performance for Soldiers study are published elsewhere [44], and only described in brief here.

The study consisted of a screening period followed by 3 separate study phases. During screening, volunteers recorded dietary intake for 3 consecutive days to assess usual energy intake. During phase 1 (study days 1–14), participants were free-living, but were provided a TBM-maintaining diet [44]. Immediately after completing phase 1, participants were admitted to the PBRC inpatient unit where they resided for 28 days during phase 2 of the study (study days 15–42). Participants were randomly assigned to one of 2 treatment groups for phase 2, testosterone (TEST), or placebo (PLA) [44]. Participants randomized to PLA received IM injections of 1 mL sesame oil on study days 15, 21, 28, and 35. Participants randomized to TEST received IM injections of 200 mg testosterone enanthate on the same days [44]. The only person not blinded to treatment allocation was the study pharmacist who prepared the treatments but did not interact with participants. During phase 2, dietary intake and physical activity were strictly monitored and controlled as described elsewhere [44], and prescribed to elicit a 55% energy deficit. Percent energy deficit did not differ between PLA (53 ± 4%) and TEST (52 ± 5%) [43].

Phase 3 (study day 43 to end-of-study [EOS]) began immediately upon completion of phase 2. Participants were released from the inpatient unit following testing on day 43, and instructed to return to their prestudy diet and exercise habits. All participants were followed for a minimum of 14 days (i.e., to study day 56). If, by day 56, a participant had returned to ±2.5% of his or her TBM measured on day 14 (as determined by average TBM over 3 consecutive days), EOS measurements were initiated and conducted over consecutive days (EOS, EOS + 1 day, EOS + 2 days). Otherwise, participants were followed until meeting the ±2.5% threshold, or until study day 84, whichever occurred first. These participants visited the laboratory weekly from day 57 to EOS for TBM measurement in addition to conducting daily TBM measurements at home. Therefore, the duration of phase 3 varied by participant and ranged from 14 to 42 days. Dietary intake during phase 3 was measured over 3 days the week before day 56 using the same methods used during screening. Study outcomes were measured during the last week of phase 1 (PRE-ED) and phase 2 (POST-ED), and at EOS as described below.

C. Appetite and Ghrelin Testing (Study Days 7, 43, and EOS + 1)

During the final week of each phase, on study days 7, 43, and EOS + 1, an experiment was conducted to determine the effects of testosterone supplementation on postprandial ghrelin and appetite responses during and after severe energy deficit (Fig. 1) [44]. Testing began at 07:00 hours on days 7 and 43, and at 07:00 hours (n = 27), 08:00 hours (n = 17), or 09:00 hours (n = 6) on EOS + 1. On each testing day, fasted participants consumed a standardized breakfast that provided energy equivalent to 20% of their individual total daily energy expenditure determined during phase 1 (PLA: 515 ± 61 kcal, TEST: 542 ± 75 kcal; P = 0.18; percent energy from carbohydrate/fat/protein = 60/25/14). After 185 minutes, a lunch meal was served (percent energy from carbohydrate/fat/protein = 41/36/23) in a portion equivalent to ≥75% of each participant’s total daily energy expenditure determined during phase 1. Participants were instructed to eat ad libitum until comfortably full, and the amount consumed was recorded. Blood samples were collected via intravenous catheter, and appetite was measured using visual analog scales 15 minutes before breakfast (served at time 0 minutes), and again 30, 60, 120, and 180 minutes after starting breakfast (Fig. 1). Appetite was measured again after lunch was finished. Visual analog scales were 100 mm in length and asked participants to rate their level of fullness, hunger, desire to eat, and the amount of food they thought they could eat (i.e., prospective consumption) at that moment [46]. Insulin, glucose, leptin, acyl ghrelin, and des-acyl ghrelin were measured at the -15 minute time point (i.e., fasting). Glucose, acyl-ghrelin, and des-acyl ghrelin concentrations were also measured at the 30, 60, 120, and 180 minute time points.

Figure 1.

Study design. Appetite and ghrelin measurements were included as secondary outcomes in the Optimizing Performance for Soldiers study [43, 44]. On study days 7, 43, and EOS + 1, appetite ratings and fasted blood samples were collected 15 minutes before a standardized breakfast, and every 30–60 minutes after starting breakfast for 180 minutes. Appetite was also measured by energy intake during a lunch meal consumed ad libitum (185 minutes). Blood samples were collected to measure fasting insulin, fasting leptin, and fasting and postprandial glucose, acyl ghrelin, and des-acyl ghrelin. EOS, end of study.

D. Biochemistries

Fasted blood samples were drawn on study days 0, 14, 42, and EOS between 06:00 hours and 09:00 hours (all day 42 measurements at 06:00 hours). Total testosterone (TT) was measured by automated immunoassay (Siemens Immulite 2000, Llanberis, UK) [47], and free testosterone (FT) was calculated according to the method of Vermeulen et al [48].

Blood samples were also drawn on appetite testing days (study days 7, 43, and EOS + 1) for measurement of fasting insulin and leptin, and fasting and postprandial glucose, acyl-ghrelin, and des-acyl ghrelin concentrations. Insulin was measured by automated immunoassay (Siemens Immulite 2000) [49]. Leptin was measured by radioimmunoassay (EMD Millipore Corporation, St Louis, MO) [50]. Glucose was measured using the Beckman DXC 600 Pro (Brea, CA). Acyl ghrelin was measured by radioimmunoassay (EMD Millipore Corporation) [51], and des-acyl ghrelin by ELISA (SPI Bio, Montigny le Bretonneux, France) [52]. All blood samples collected for measurement of acyl ghrelin were collected into chilled tubes containing 1 mg/mL 4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride and were kept on ice until processing. After centrifugation, plasma acyl ghrelin aliquots were acidified using HCl to a final concentration of 0.05N. Des-acyl ghrelin was collected into chilled EDTA tubes and kept on ice until centrifugation. Samples were stored at -80°C until analysis. Total ghrelin concentration was calculated as the sum of acyl and des-acyl ghrelin.

E. Anthropometrics and Body Composition

Semi-nude, fasted, morning TBM was measured during laboratory visits using a calibrated digital scale (GSE Inc. Model 450, GSE Scale Systems, Novi, MI). Participants weighed themselves daily during phase 3 when they did not visit the laboratory using a calibrated scaled provided by research staff (Body Trace, Inc. Model BT003, New York, NY). TBM, FM, and bone mass were determined by dual-energy X-ray absorptiometry (Lunar iDXA, GE Healthcare, Madison, WI) on study days 0, 11, 39, and EOS after an overnight fast. Lean body mass (LBM) was defined as TBM minus FM and bone mass.

F. Statistical Analysis

As described elsewhere [43, 44], sample size estimates were based on the primary study outcome, LBM, and indicated that 22 participants/group would provide 90% power to detect an expected 25% attenuation in LBM loss with testosterone supplementation. Sample size estimates for postprandial ghrelin area under the curve (AUC) were based on an expected 25% between-group difference following testosterone supplementation [36]. Using means and standard deviations from previous work by our laboratory conducted in a similar population [53], we calculated that an effect of that magnitude could be detected with 19 participants/group with 80% power and α = 0.05.

All data were examined quantitatively and graphically, and logarithmic transformations were applied when necessary to meet model assumptions. Between-group comparisons of preintervention measurements were conducted using Student t-test or the Mann-Whitney U-test as appropriate. Linear mixed models with participant included as a random factor were used to examine effects of treatment, study day, and their interaction on fasting appetite ratings, glucose and hormone concentrations, body composition, and ad libitum energy intake at lunch. Models for testosterone and body composition were adjusted for day 0 values of each variable. The AUC with respect to ground (y = 0) was calculated for postprandial appetite ratings, and postprandial glucose and ghrelin concentrations using the trapezoidal method [54]. AUC was analyzed using linear mixed models as described above. Significant main effects and interactions were investigated further using between-group and within-group pairwise comparisons. Bonferroni corrections were applied to adjust for multiple comparisons. Because of evidence for between group differences in PRE-ED ghrelin concentrations (see Results), mixed models with PRE-ED values included as a covariate were also used to analyze between-group differences in fasting ghrelin concentrations and postprandial ghrelin AUC. Finally, in exploratory analyses, correlations between study outcomes were assessed using Pearson or Spearman rank correlations as appropriate, and multiple linear regression was used to assess relationships between variables while controlling for covariates.

Statistical analyses were completed using SPSS v24.0 (IBM, Armonk, NY). Tests were 2-sided and considered statistically significant at P ≤ 0.05. Results are presented as mean ± SD or adjusted least square mean difference (95% confidence interval [CI]) unless otherwise noted.

2. Results

Of the 53 men who enrolled, 50 were randomized (60% non-Hispanic white, 26% non-Hispanic black, 6% Hispanic, 8% other) and all 50 completed the study [43]. Changes in testosterone concentrations and body composition measured between study days 0 to 56 have been previously reported [43]. To provide context for this secondary analysis, these changes are presented in Table 1 and briefly summarized here. EOS values have not been previously reported.

Table 1.

Body Composition and Fasting Testosterone Concentrations

	PLA (n = 26)			TEST (n = 24)			P Value
	PRE-ED (Day 11 or 14)	POST-ED (Day 39 or 42)	EOS	PRE-ED (Day 11or 14)	POST-ED (Day 39 or 42)	EOS	Phase	Group	P × G
Body mass (kg)1	76.3 ± 10.2	71.3 ± 9.8#	74.9 ± 9.0#‡	80.2 ± 13.5	78.0 ± 12.3#*	81.0 ± 12.6‡*	<0.001	<0.001	<0.001
Fat mass (kg)1§	17.0 ± 6.7	12.3 ± 6.6	14.0 ± 5.4	16.6 ± 6.9	11.8 ± 6.1	13.7 ± 5.3	<0.001	0.67	0.93
Lean mass (kg)1	56.1 ± 5.1	55.8 ± 5.0	57.7 ± 5.1#‡	60.3 ± 10.1	62.8 ± 9.5#*	64.0 ± 10.2#‡*	<0.001	<0.001	<0.001
Total testosterone (ng/dL)2	402 ± 101	319 ± 131	424 ± 120‡	472 ± 153	1016 ± 246#*	324 ± 135#‡*	<0.001	<0.001	<0.001
Free testosterone (ng/dL)2,3	8.8 ± 2.3	5.0 ± 2.5#	8.8 ± 2.7‡	9.7 ± 2.4	24.4 ± 8.0#*	6.4 ± 2.7#‡*	<0.001	<0.001	<0.001

ED, energy deficit; EOS, end of study (return to initial body mass); PLA, placebo treatment group; TEST, testosterone treatment group.

Values are mean ± SD. Analyzed by linear mixed models with Bonferroni corrections, and adjusted using day 0 values.

*Significant difference from PLA at same time point (P < 0.05).

^#Significant difference from PRE-ED within the same treatment group (P < 0.05).

^‡Significant difference from POST-ED within the same treatment group (P < 0.05).

^§Main effect of phase, all phases are significantly different (P < 0.05).

¹Measured on study days 11 (phase 1), 39 (phase 2) and EOS.

²Measured on study days 14 (phase 1), 42 (phase 2) and EOS.

³Log₁₀-transformed for analysis.

Within PLA, FT and TT concentrations were stable or declined from PRE-ED to POST-ED, then returned to PRE-ED concentrations by EOS; whereas, within TEST, FT and TT concentrations increased from PRE-ED to POST-ED, then decreased to below PRE-ED concentrations by EOS. Relative to PLA, concentrations of both TT and FT were higher in TEST at POST-ED but lower at EOS (Table 1). Changes in FM did not differ between groups (Table 1). In contrast, TBM loss from PRE-ED to POST-ED was attenuated in TEST relative to PLA because of an increase in LBM in TEST (Table 1). Participants in TEST (median [interquartile range] = 2 days [3]) then returned to their initial TBM earlier than those in PLA (11 days [16], P = 0.001). Five participants in PLA and 1 in TEST (Pearson χ ² = 2.7, P = 0.10) did not return to initial TBM within 42 days of ad libitum eating, which was the predetermined EOS for participants who did not fully recover lost TBM. As a result, EOS TBM in PLA remained 1.4 kg (95% CI: 0.4-2.5) lower than initial TBM (P = 0.005), but did not differ between the same time point in TEST (0.8 kg [95% CI: -0.3 to 1.9] P = 0.22).

A. Glucose and Hormone Concentrations

Testosterone supplementation had no effect on fasting leptin and insulin concentrations which were both lower at POST-ED relative to other time points (Table 2). Fasting glucose concentrations decreased in both groups from PRE-ED to POST-ED and were higher at EOS relative to PRE-ED in TEST but not PLA (Table 2). Testosterone supplementation had no effect on postprandial glucose AUC (data not shown).

Table 2.

Fasting Glucose, Insulin, Leptin and Ghrelin Concentrations, Fasting Appetite Ratings, and Energy Intake During an ad libitum Lunch Meal

	PLA (n = 26)			TEST (n = 24)			P Value
	PRE-ED (Day 7)	POST-ED (Day 43)	EOS + 1 day	PRE-ED (Day 7)	POST-ED (Day 43)	EOS + 1 day	Phase	Group	P × G
Glucose (mg/dL)	86 ± 7	79 ± 6#	86 ± 7‡	85 ± 7	76 ± 6#	90 ± 6#‡	<0.001	0.98	0.01
Insulin (μIU/mL)1†	4.5 [6.3]	0.2 [2.7]	5.7 [6.4]	4.6 [3.8]	0.2 [2.1]	6.3 [5.9]	<0.001	0.56	0.97
Leptin (ng/mL)1†	6.2 [6.1]	3.2 [1.7]	5.8 [6]	5.4 [7.9]	3.2 [1.2]	4.8 [3.1]	<0.001	0.65	0.77
Acyl ghrelin (pg/mL)	118 ± 44	145 ± 59#	130 ± 49	137 ± 53	132 ± 59	141 ± 52	0.27	0.65	0.05
Des-acyl ghrelin (pg/mL)1†	163 [108]	204 [121]	199 [102]	228 [103]	244 [177]	228 [172]	<0.001	0.38	0.68
Total ghrelin (pg/mL)1¶	287 [168]	326 [233]	310 [160]	350 [180]	353 [273]	357 [276]	0.002	0.49	0.15
Acyl/des-acyl ghrelin ratio	0.65 ± 0.20	0.64 ± 0.25	0.71 ± 0.36	0.64 ± 0.30	0.49 ± 0.22	0.59 ± 0.21	0.08	0.49	0.15
Fullness (mm)1†	13 [33]	9 [23]	12 [34]	15 [39]	10 [32]	17 [34]	0.01	0.91	0.60
Hunger (mm)¥	59 ± 24	73 ± 16	68 ± 22	62 ± 21	75 ± 18	73 ± 15	<0.001	0.46	0.85
PC (mm)	66 ± 19	74 ± 20	70 ± 21	69 ± 19	75 ± 17	73 ± 19	0.12	0.75	0.81
DtE (mm)†	62 ± 26	76 ± 20	68 ± 23	66 ± 25	77 ± 19	70 ± 22	0.001	0.64	0.86
Energy intake at lunch meal (kcal)†	1211 ± 438	1418 ± 560	1108 ± 455	1145 ± 366	1581 ± 544	1241 ± 468	<0.001	0.52	0.11

DtE, Desire to eat; ED, energy deficit; EOS, end of study (return to initial body mass); PLA, placebo treatment group; PC, prospective consumption; TEST, testosterone treatment group.

Values are mean ± SD or median [IQR]. Analyzed by linear mixed models with Bonferroni corrections.

*Significant difference from PLA at same time point (P < 0.05).

^#Significant difference from PRE-ED within the same treatment group (P < 0.05).

^‡Significant difference from POST-ED within the same treatment group (P < 0.05).

^†Main effect of phase, POST-ED differs from PRE-ED and EOS.

^¶Main effect of phase, POST-ED differs from PRE-ED only.

^¥Main effect of phase, PRE-ED differs from POST-ED and EOS.

¹Log₁₀-transformed for analysis.

Fasting acyl ghrelin concentrations were not statistically different between groups before treatment (i.e., at PRE-ED; P = 0.16). However, statistical adjustment for PRE-ED acyl ghrelin affected model results. When PRE-ED concentrations were not included in the model, a phase-by-group interaction was observed (P = 0.05). Fasting acyl ghrelin was found to increase from PRE-ED to POST-ED and then return to PRE-ED values at EOS in PLA but did not change over time in TEST (Table 2). In contrast, after statistically adjusting for PRE-ED concentrations, no significant treatment effects were observed (phase-by-group interaction, P = 0.11).

Postprandial acyl ghrelin AUC was higher in TEST relative to PLA before treatment (P = 0.05; Fig. 2C). However, statistical adjustment for PRE-ED acyl ghrelin AUC did not affect model results. When PRE-ED AUC was not included in the model, a phase-by-group interaction was observed (P = 0.004). Acyl ghrelin AUC increased from PRE-ED to POST-ED and then returned to PRE-ED values at EOS in PLA, but did not change over time in TEST (Fig. 2A-C). The phase-by-time interaction remained significant after adjusting for PRE-ED acyl ghrelin AUC (phase-by-group interaction, P = 0.02). Following the adjustment, POST-ED acyl ghrelin AUC was 21% higher in PLA relative to TEST (adjusted mean difference = 4189 pg × min/mL [95% CI: 1679-6699] P = 0.001), and no between group difference was observed at EOS (P = 0.75). Finally, when postprandial AUC was recalculated as a change from fasting concentrations, no effect of testosterone was observed (data not shown; phase-by-group interaction, P = 0.58), indicating that the effect of testosterone on acyl ghrelin AUC was primarily attributable to a change in fasting concentrations rather than a change in the magnitude of the postprandial response.

Figure 2.

Testosterone supplementation alters postprandial ghrelin responses during energy deficit. Fasting and postprandial acyl ghrelin (A-C), des-acyl ghrelin (D-F), total ghrelin (G-I), and the acyl ghrelin/des-acyl ghrelin ratio (J-L) measured during a standardized meal challenge (0–180 minutes) administered before (PRE; study day 7) and after (POST; study day 43) severe energy deficit (ED), and 1 day after body mass recovery (EOS). (A-B, D-E, G-H, J-K) Individual data points are mean ± SEM. (C, F, I, L) Boxes span 25th to 75th percentile with the median value represented by a solid line inside the box. Whiskers extend to minimum and maximum values or to 1.5 times the interquartile range if any values outside that range (represented by circles). Main effects of group, study phase, and their interaction on area under the postprandial response curve (AUC) analyzed by linear mixed models with Bonferroni corrections. (C, F, I) Group-by-phase interaction, P < 0.05. (L) Main effect of group, P = 0.05. (C, F, I, L) * P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. PLA, placebo treatment group; TEST, testosterone treatment group.

Fasting des-acyl ghrelin concentrations trended toward being statistically different between groups before treatment (P = 0.06). In the full cohort, fasting des-acyl ghrelin concentrations increased from PRE-ED to POST-ED and then decreased and did not differ from PRE-ED concentrations at EOS (Table 2). No effect of testosterone supplementation was observed whether or not PRE-ED des-acyl ghrelin was included in the model.

Postprandial des-acyl ghrelin AUC was higher in TEST relative to PLA before treatment (P = 0.01). However, statistical adjustment for PRE-ED des-acyl ghrelin AUC did not affect model results. When PRE-ED AUC was not included in the model, a phase-by-group interaction was observed (P = 0.02). Des-acyl ghrelin AUC increased from PRE-ED to POST-ED and then returned to PRE-ED values at EOS in PLA but did not change over time in TEST (Fig. 2D-F). The phase-by-time interaction remained significant after adjusting for the PRE-ED differences in des-acyl ghrelin AUC (phase-by-group interaction, P = 0.05). The ratio of the geometric mean of POST-ED des-acyl ghrelin AUC in PLA relative to that in TEST was 1.15 (95% CI: 1.03, 1.29; P = 0.01), and no between-group difference was observed at EOS (P = 0.98; phase-by-group interaction, P = 0.02). When postprandial AUC was recalculated as a change from fasting des-acyl ghrelin concentrations, no effect of testosterone was observed (data not shown; phase-by-group interaction, P = 0.97), suggesting that the effect of testosterone on des-acyl ghrelin AUC was attributable to a combination of changes in fasting concentrations and in the magnitude of the postprandial response.

Results for fasting total ghrelin and total ghrelin AUC largely mirrored those for des-acyl ghrelin which was the predominant form in circulation (Table 2andFig. 2G-I). Testosterone supplementation had no effect on the ratio of acyl to des-acyl ghrelin under either fasting or postprandial conditions (Table 2andFig. 2J-L).

B. Appetite

Fasting and postprandial appetite ratings demonstrated an increase in hunger, desire to eat, prospective consumption, and a decrease in fullness from PRE-ED to POST-ED that was fully or partially resolved by EOS and not affected by testosterone supplementation (Table 2, Fig. 3). Changes in energy intake during the ad libitum lunch mirrored that response and did not differ by study group (Table 2). The difference in self-reported energy intake measured during the first week of phase 3 relative to that measured during screening also did not differ between groups (PLA: 347 kcal/d [95% CI: -171 to 864] vs. TEST: 900 kcal/d [95% CI: 432-1368], P = 0.11). However, the median [IQR] rate of TBM regain from POST-ED until a participant was considered returned to their initial TBM was higher in TEST (1.7 kg/d [1.9]) relative to PLA (0.4 kg/d [1.1]; P = 0.03).

Figure 3.

Testosterone supplementation does not affect appetite during energy deficit. Fasting and postprandial fullness (A-C), hunger (D-F), prospective consumption (Pro. Consumption, G-I), and desire to eat (J-L) measured during a standardized breakfast (0–180 minutes) followed by an ad libitum lunch meal (served at 185 minutes) administered before (PRE; study day 7) and after (POST; study day 43) severe energy deficit (ED), and one day after body mass recovery (EOS). (A-B, D-E, G-H, J-K) Individual data points are mean ± SEM. (C, F, I, L) Boxes span 25th to 75th percentile with the median value represented by a solid line inside the box. Whiskers extend to minimum and maximum values or to 1.5 times the interquartile range if any values outside that range (represented by circles). Main effects of group, study phase, and their interaction on area under the postprandial response curve (AUC) analyzed by linear mixed models with Bonferroni corrections. (C, F, I, L) Main effect of phase, P < 0.05. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. PLA, placebo treatment group; TEST, testosterone treatment group.

C. Associations Between Changes in Testosterone, Ghrelin, and Appetite

In the full cohort, changes in TT and FT concentrations were inversely correlated with changes in fasting acyl ghrelin, but not fasting des-acyl ghrelin, concentrations (Table 3). The correlations between changes in FT concentrations and changes in fasting acyl ghrelin concentrations remained statistically significant after adjusting for changes in FM, LBM, and TBM (Table 4).

Table 3.

Correlations Between Changes in Fasting Ghrelin Concentrations and Changes in Testosterone and Body Mass Outcomes

	∆daG	∆totG	∆aG/daG	∆TT	∆FT	∆FM	∆LBM	∆TBM
∆PRE-ED to POST-ED
∆acyl ghrelin	0.46***	0.69***	0.40**	-0.37**	-0.42**	0.12	-0.12	-0.04
∆des-acyl ghrelin		0.88***	-0.48***	-0.03	-0.04	0.16	-0.05	0.01
∆total ghrelin			-0.12	-0.23	-0.23	0.20	-0.11	-0.03
∆acyl/des-acyl ghrelin				-0.26	-0.34*	-0.03	-0.11	-0.13
∆POST-ED to EOS
∆acyl ghrelin	0.41**	0.75***	0.45***	-0.39**	-0.38**	-0.42**	-0.15	-0.31*
∆des-acyl ghrelin		0.86***	-0.45***	-0.14	-0.12	-0.36*	0.01	-0.18
∆total ghrelin			-0.04	-0.31*	-0.30*	-0.41**	-0.06	-0.27
∆acyl/des-acyl ghrelin				-0.17	-0.22	-0.09	-0.13	-0.15

n = 49–50. ∆change; aG, acyl ghrelin; daG; des-acyl ghrelin; FM, fat mass; FT, free testosterone; LBM, lean body mass; TBM, total body mass; TT, total testosterone. Values are Pearson’s correlation coefficients or Spearman’s rho. All correlations were computed using change scores calculated over the same time period for the independent and dependent variables (ie, PRE-ED to POST-ED and POST-ED to EOS).

*P < 0.05; **P ≤ 0.01; ***P ≤ 0.001.

Table 4.

Association Between Changes in Fasting Acyl Ghrelin Concentrations and Changes in Free Testosterone (FT) Concentrations are not Mediated by Changes in Body Composition or Body Mass

	Model 1a		Model 2		Model 3		Model 4
	β [95% CI]	P	β [95% CI]	P	β [95% CI]	P	β [95% CI]	P
∆PRE-ED to POST-ED
∆FT (ng/dL)	-1.7 [-3.0, -0.4]	0.01	-1.7 [-3.1, -0.4]	0.01	-2.6 [-4.6, -0.6]	0.01	-2.7 [-4.6, -0.8]	0.01
Age (yr)	-2.5 [-5.6, 0.7]	0.12	-2.4 [-5.6, 0.7]	0.13	-2.3 [-5.5, 0.8]	0.14	-2.3 [-5.4, 0.9]	0.15
Baseline BF (%)	0.3 [-2.1, 2.7]	0.78	0.7 [-2.2, 3.7]	0.61	0.2 [-2.2, 2.6]	0.84	0.9 [-1.6, 3.5]	0.45
∆FM (kg)	-		3.9 [-12.0, 20.0]	0.62	-		-
∆LBM (kg)	-		-		6.3 [-4.4, 1.7]	0.24	-
∆TBM (kg)	-		-		-		6.8 [-2.9, 16.5]	0.16
Constant	74.3 [-2.8, 151.5]	0.06	83.2 [-2.5, 168.9]	0.06	70.8 [-6.2, 147.9]	0.07	85.8 [7.7, 163.8]	0.03
R2	0.20	0.02	0.20	0.04	0.22	0.02	0.23	0.02
∆POST-ED to EOS
∆FT (ng/dL)	-1.5 [-2.6, -0.3]	0.01	-1.4 [-2.5, -0.2]	0.02	-1.6 [-2.9, -0.4]	0.01	-1.5 [-2.7, -0.2]	0.02
Age (yr)	0.8 [-2.3, 3.9]	0.59	0.2 [-3.1, 3.5]	0.90	0.7 [-2.5, 3.9]	0.67	0.8 [-2.3, 4.0]	0.59
Baseline BF (%)	3.0 [0.6, 5.3]	0.01	2.0 [-0.9, 4.8]	0.16	3.1 [0.7, 5.6]	0.01	2.9 [0.3, 5.6]	0.03
∆FM (kg)	-		9.7 [-2.6, 6.4]	0.23	-		-
∆LBM (kg)	-		-		2.5 [-6.4, 1.1]	0.57	-
∆TBM (kg)	-		-		-		-0.2 [-7.1, 6.7]	0.95
Constant	-97.1 [-173.3, -20.9]	0.01	-42.7 [-160.6, 75.3]	0.47	-101.5 [-179.9, -23.1]	0.01	-95.6 [-187.8, -3.3]	0.04
R2	0.28	0.002	0.30	0.003	0.29	0.004	0.28	0.005

n = 49. ∆, change; BF, body fat; FM, fat mass; LBM, lean body mass; TBM, total body mass.

^aAll models are multiple linear regression analyses. Dependent variable is change in fasting acyl ghrelin concentration over the time period indicated in the table.

Changes in postprandial acyl ghrelin, des-acyl ghrelin, total ghrelin, and the acyl/des-acyl ghrelin ratio AUCs from PRE-ED to POST-ED were not correlated with changes in postprandial hunger AUC during the same period, or with changes in ad libitum energy intake as measured by: 1) the difference in self-reported energy intake during the first week of phase 3 relative to screening; 2) the difference in lunch meal ad libitum energy intake at POST-ED relative to PRE-ED; or 3) the rate of TBM regain from POST-ED until initial TBM was reached (data not shown).

3. Discussion

In this secondary analysis, testosterone supplementation during severe energy deficit prevented deficit-induced increases in fasting acyl ghrelin concentrations, and in postprandial acyl and des-acyl ghrelin concentrations in nonobese, healthy men. Despite an established role for ghrelin in energy balance regulation, these effects did not blunt energy deficit-induced increases in appetite, slow TBM recovery, or impact changes in FM. These findings confirm and extend evidence supporting an inhibitory effect of testosterone supplementation on circulating ghrelin in healthy men, but suggest this effect is not sufficient to suppress appetite during energy deficit or subsequent body mass recovery.

Whereas previous studies have examined associations between testosterone supplementation and fasting total ghrelin concentrations [36, 37, 41, 42], to our knowledge, this study is the first to examine relationships between testosterone and both fasting and postprandial acyl and des-acyl ghrelin forms. This is a strength of the study, and important to developing a complete understanding of any effects of testosterone supplementation on ghrelin because: 1) acyl ghrelin is considered the more biologically active ghrelin form [26]; 2) acyl and des-acyl ghrelin may have independent, and, in some cases, opposing effects [32]; and 3) most individuals spend the majority of the day in a postprandial state [55, 56]. Using this comprehensive approach, this study extends the findings of Gambineri et al [36], who demonstrated that supplementing testosterone to supraphysiologic concentrations (~1400ng/dL; ~60% increase from baseline) resulted in a ~25% decrease in fasting total ghrelin concentrations over 8wk in normal-weight men in whom endogenous testosterone production was suppressed by oral progestin administration. The present study demonstrates that those results may be driven more by changes in acyl ghrelin than des-acyl ghrelin, and that the effects on acyl ghrelin concentrations can be produced at testosterone concentrations that span the lower to upper limits of the normal physiologic range (300-1000 ng/dL [57]; 115% increase from baseline). Further, these findings demonstrate that testosterone supplementation does not appear to measurably impact the magnitude of the postprandial acyl or des-acyl ghrelin response independent of changes in fasting concentrations (Table 2 and Fig. 2).

Changes in TBM, FM, and LBM did not mediate associations between testosterone and fasting acyl ghrelin concentrations (Table 4). This suggests that the observed effects of testosterone on acyl ghrelin were independent of changes in TBM and body composition, both of which have previously been associated with ghrelin concentrations [26]. The result contrasts with a recent study in which decreased TBM and FM were hypothesized to override any effect of testosterone treatment on fasting total ghrelin concentrations in men with obesity and low-normal testosterone concentrations who were consuming a very low energy diet [42]. Whether the discrepancy is due to the absence of acyl ghrelin measurements in that study, or due to differences in the study populations, the magnitude of TBM and FM loss or the method used to induce energy deficit (increased energy expenditure vs. very low energy diet) is unclear. Additionally, it has been postulated that any effect of testosterone on ghrelin concentrations may be indirect and mediated through changes in insulin or leptin concentrations [58]. However, neither insulin nor leptin was affected by testosterone supplementation in this study. Of note, multiple endocrine, neural, and nutritive factors collectively regulate basal and periprandial ghrelin concentrations, and the combinations of those factors underpinning ghrelin responses to body mass loss are not fully characterized [26, 58]. Study findings strengthen evidence that testosterone may be 1 of those factors. Conversely, ghrelin has been shown to suppress hypothalamic-pituitary-gonadal axis activity [59, 60], and an energy deficit-induced increase in ghrelin is thought to be 1 mechanism by which reproductive function is suppressed when energy availability is low [61, 62]. Taken together, these observations add support for a bidirectional interaction between testosterone and ghrelin linking energy homeostasis and male reproductive function.

Despite preventing energy deficit-induced increases in acyl ghrelin, testosterone supplementation did not blunt deficit-induced increases in appetite, or slow recovery of TBM and FM during refeeding. These results are incongruent with acyl ghrelin’s putative role as an orexigenic and adipogenic hormone. However, clinical studies demonstrating acute orexigenic effects of ghrelin have generally used IV infusions resulting in supraphysiologic acyl and total ghrelin concentrations [63]. Studies conducted in normal-weight adults using lower dose infusions that produce more physiologically relevant ghrelin concentrations have not reported statistically significant increases in appetite despite 1.6- to 2.4-fold increases in circulating acyl and total ghrelin [64, 65]. Further, some [66–68], but not all [30], rodent studies suggest that des-acyl ghrelin may antagonize the appetite-stimulating effects of acyl ghrelin. We therefore posit that the ~1.25-fold between-group difference in changes in fasting acyl-ghrelin concentrations in this study were likely not sufficient to result in between-group differences in hunger, energy intake or changes in FM in the absence of differences in the acyl:des-acyl ghrelin ratio. Additionally, eating behavior is influenced by myriad interactions among endocrine, neural, and other factors [69] that are sensitive to energy deficit, and promote an appetitive drive to regain lost body mass [22]. Responses of those factors to energy deficit may have overwhelmed any influence of testosterone-mediated differences in ghrelin on appetite.

Strengths of this study include the lack of attrition, strict control of energy intake and expenditure, assessment of postprandial acyl and des-acyl ghrelin dynamics, and concomitant measurements of appetite. However, results should be interpreted within the context of several limitations. First, as ghrelin and appetite were secondary outcomes, the trial was not designed specifically to examine interactions between those outcomes and testosterone. Second, despite randomization, between-group differences in acyl ghrelin and des-acyl ghrelin AUC were noted before testosterone treatment. However, when statistical adjustment for the pretreatment differences was used, both acyl ghrelin AUC and des-acyl ghrelin AUC were higher at the POST-ED time point in PLA relative to TEST. Further, in confirmatory analysis of covariance analyses adjusting for PRE-ED ghrelin AUC, changes in acyl and des-acyl ghrelin AUCs from PRE-ED to POST-ED remained greater in PLA relative to TEST (data not presented). Taken together, these results are consistent with a testosterone-mediated suppression of ghrelin. Because no statistically significant between-group differences in changes in fasting ghrelin concentrations were observed, these findings also highlight the value of obtaining serial measurements for appetite-associated hormones that deviate from fasting concentrations throughout much of the day.

An additional study limitation is that mean TT concentrations in PLA did not fall as precipitously as has been reported in studies of military personnel experiencing severe energy deficit, and the substantial LBM losses reported in those studies were not reproduced [6, 12]. The presence of additional physiologic stressors in previous studies, such as sleep deprivation, likely explain these differences. However, we previously reported that PLA did experience reductions in luteinizing hormone in addition to reductions in FT, and increases in sex-hormone binding globulin during energy deficit [43]. Those responses are consistent with hypothalamic-pituitary-gonadal axis suppression [57]. Therefore, study findings can to some extent be extrapolated to the real-life conditions the study was intended to mimic. Results may also be generalizable to physiologic adaptations experienced in men in whom energy imbalances result from intentional or unintentional restriction of energy intake and increases in energy expenditure during athletic training [61]. Finally, mean POST-ED testosterone concentrations in TEST were at the high end of the normal physiologic range for young men and substantially higher than PRE-ED concentrations. This allowed an examination of relationships between testosterone, ghrelin, and appetite across the full range of normal physiologic concentrations. However, whether similar effects would have been observed had testosterone concentrations been maintained closer to the normal basal range in this population cannot be determined.

4. Conclusions

Testosterone supplementation blunted energy deficit-induced increases in ghrelin, associations between changes in testosterone and ghrelin were not mediated by changes in body composition, and effects of testosterone supplementation on ghrelin did not suppress appetite. Testosterone supplementation therefore appears unlikely to appreciably influence appetite or energy intake in nonobese young men, to include military personnel, during and following periods of unavoidable severe energy deficit. However, observations strengthen evidence of testosterone-ghrelin interactions, and provide additional insight into endocrine factors potentially linking energy homeostasis and male reproductive function.

Abbreviations

Abbreviations
 
• AUC

  area under the curve

 
• CI

  confidence interval

 
• ED

  energy deficit

 
• EOS

  end-of-study time point

 
• FM

  fat mass

 
• FT

  free testosterone

 
• LBM

  lean body mass

 
• PLA

  placebo group

 
• POST-ED

  posttreatment time point

 
• PRE-ED

  pretreatment time point

 
• TBM

  total body mass

 
• TEST

  testosterone treatment group

 
• TT

  total testosterone

Acknowledgments

We thank the study participants who participated in the study, the study team at the PBRC for significant contributions to study management and data collection, Dr. Shalender Bhasin for assistance in designing the testosterone intervention, and Philip Niro for assistance in designing the figures.

Financial Support: This work was funded by the US Department of Defense under the Collaborative Research to Optimize Warfighter Nutrition II and III projects and the Joint Program Committee-5. C.E.B. was supported in part by an appointment to the Department of Defense (DoD) Research Participation Program administered by the Oak Ridge Institute for Science and Education (ORISE) through an interagency agreement between the U.S. Department of Energy (DoE) and the DoD. ORISE is managed by the Oak Ridge Associated Universities (ORAU) under DoE contract number DE-SC0014664.

Additional Information

Disclosure Statement: All opinions or assertions contained herein are the private views of the authors and are not to be construed as official or reflecting the views of the U.S. Army, DoD, DoE, or ORAU/ORISE. Any citations of commercial organizations and trade names in this report do not constitute an official Department of the Army endorsement or approval of the products or services of these organizations. Approved for public release; distribution is unlimited.