Testosterone Suppresses Hepcidin in Men: A Potential Mechanism for Testosterone-Induced Erythrocytosis

Abstract

Context: The mechanisms by which testosterone increases hemoglobin and hematocrit are unknown.

Objective: The aim was to test the hypothesis that testosterone-induced increase in hematocrit is associated with suppression of the iron regulatory peptide hepcidin.

Participants: Healthy younger men (ages 19–35 yr; n = 53) and older men (ages 59–75 yr; n = 56) were studied.

Methods: Weekly doses of testosterone enanthate (25, 50, 125, 300, and 600 mg) were administered over 20 wk, whereas endogenous testosterone was suppressed by monthly GnRH agonist administration. Blood and serum parameters from each individual were measured at wk 0, 1, 2, 4, 8, and 20. Longitudinal analyses were performed to examine the relationship between hepcidin, hemoglobin, hematocrit, and testosterone while controlling for potential confounders.

Results: High levels of testosterone markedly suppressed serum hepcidin within 1 wk. Hepcidin suppression in response to testosterone administration was dose-dependent in older men and more pronounced than in young men, and this corresponded to a greater rise in hemoglobin in older men. Serum hepcidin levels at 4 and 8 wk were predictive of change in hematocrit from baseline to peak levels.

Conclusion: Testosterone administration is associated with suppression of serum hepcidin. Greater increases in hematocrit in older men during testosterone therapy are related to greater suppression of hepcidin.

Testosterone suppresses the iron regulatory peptide hepcidin in men, and this effect is more pronounced in older men who also experience greater erythrocytosis.

Testosterone regulates erythropoiesis in numerous mammalian species, including humans of both sexes (1). Excessive erythrocytosis is the most common serious adverse event associated with testosterone therapy in older men (2). However, the mechanisms by which testosterone stimulates erythropoiesis remain poorly understood. It has been suggested that testosterone stimulates erythropoietin secretion and directly stimulates erythroid progenitor cells (3,4). We showed previously, however, that testosterone dose-dependently increases hemoglobin and hematocrit, but without an associated increase in erythropoietin (5). In addition, testosterone has minimal proliferative effect on purified (CD34+) erythroid progenitors ex vivo (6). We considered the hypothesis that testosterone increases hematocrit by suppressing the master iron regulatory peptide hepcidin, thus resulting in increased bioavailable iron. Hepcidin is a liver-derived peptide that binds to and degrades the iron channel ferroportin (7,8). Increased hepcidin, in response to infection and inflammation restricts systemic iron bioavailability and results in mild anemia in chronic disease (9). Low hepcidin, conversely, is associated with increased iron absorption, increased systemic iron transport, and erythropoiesis. To test the hypothesis that testosterone suppresses serum hepcidin, we measured serum hepcidin levels in a testosterone dose response study, in which healthy younger (19–35 yr old) and older (59–75 yr) men were administered a long-acting GnRH agonist to suppress endogenous testosterone production, along with varying doses of testosterone enanthate for 20 wk (10,11). This design produced cohorts of subjects with graded, stable levels of testosterone within 4 wk that were maintained for 20 wk. This intervention resulted in dose-dependent increases in hematocrit and hemoglobin that were greater in older than younger men (5). We measured serum hepcidin in serum samples from these men, and tested the hypothesis that age-related differences in erythropoietic response are related to the magnitude of hepcidin suppression. We also assessed whether early changes in hepcidin levels predict subsequent changes in hematocrit and hemoglobin.

Subjects and Methods

Serum samples from a prior testosterone dose-response study were analyzed (10,11). These trials were randomized, double-blinded studies that consisted of a 4-wk control period, a 20-wk testosterone treatment period, and 16 wk of recovery. All subjects received monthly injections of GnRH agonist (leuprolide; TAP Pharmaceutical Products, North Chicago, IL) to suppress endogenous testosterone production, in addition to weekly doses of 25, 50, 125, 300, or 600 mg testosterone enanthate for a total of 20 wk (5). Testosterone, free testosterone, hematocrit, hemoglobin, IL-6, and C-reactive protein (CRP) levels and serum erythropoietin were measured by previously described methods (5). Hepcidin was measured by ELISA (Intrinsic LifeSciences, La Jolla, CA) as previously reported (12). Hepcidin peptide has been shown to remain stable after repeated freeze-thaw cycles, and consistent reference ranges have been reported (12,13). Ferritin, serum iron, iron binding capacity, and percentage saturation were measured at Quest Diagnostics (Cambridge, MA). Summary statistics were calculated to inspect the differences in the baseline parameters as well as changes in their response to testosterone treatment between younger and older men. Correlation statistics were computed to assess the unadjusted associations between variables under the assumption of linear relationships and were stratified by study/age groups. Exploratory analyses using generalized additive models were employed; linear mixed effect models (14) with random subject intercepts were used to test whether the changes in hepcidin concentrations were associated with testosterone, hematocrit, and hemoglobin. Analyses were adjusted for age and interactions with other predictors. Statistical significance was determined using a type-I error probability of 0.05, using R version 2.9.2 (R Foundation for Statistical Computing, Vienna, Austria) and SAS 9.2 (SAS Institute, Cary, NC).

Results

Baseline characteristics

Older men had higher body mass index (BMI) than younger men, mostly due to higher total body fat (Table 1). Lean mass was similar in older vs. younger men. Baseline free and total testosterone levels were within the normal range for men, and were significantly higher in young men than older men (10,11). Baseline hemoglobin and hematocrit levels were similar in young men compared with older men. Compared with young men, older men had significantly higher serum hepcidin levels by almost 2-fold, even after adjusting for baseline testosterone levels, although there was considerable interindividual variation as previously reported (12,13). Serum ferritin levels were also significantly higher in older men. Serum levels of IL-6 and serum iron (Fe) tended to be higher in older men, but this did not reach statistical significance.

Table 1.

Baseline characteristics of study participants

	Younger	Older	P value
n	53	56
General variables
Age (yr)	26.4 ± 4.3	65.6 ± 4.3	NA
BMI (kg/m2)	24.1 ± 3.0	26.7 ± 3.3	0.0001
Fat mass (kg)	13.9 ± 6.4	22.1 ± 6.8	2.0 × 10−8
Lean mass (kg)	57.4 ± 7.1	57.9 ± 6.4	0.7494
Serum
Hepcidin (ng/ml)	63.9 ± 54.0	112.6 ± 69.5	0.0004
Free testosterone (ng/dl)	6.1 ± 1.8	3.4 ± 0.9	7.2 × 10−12
Total testosterone (ng/dl)	618.7 ± 163.7	354.4 ± 88.3	3.5 × 10−12
IL-6 (pg/ml)	1.7 ± 2.2	3.1 ± 5.6	0.0803
Fe (μg/dl)	275.7 ± 139.8	293.1 ± 150.1	0.5967
Ferritin (ng/ml)	102.2 ± 64.6	181.4 ± 168.6	0.0094
Hematocrit (%)	44.0 ± 2.2	43.0 ± 3.1	0.0560
Hemoglobin (g/liter)	145.5 ± 12.8	146.1 ± 10.7	0.7790

Data are expressed as mean ± sd. P values were from two-sample t tests comparing the younger and older groups. BMI, fat, lean mass, testosterone, hematocrit, and hemoglobin were reported previously (5,10,11). Baseline hepcidin and serum ferritin were significantly higher in older compared to younger men. NA, Not available. 

Testosterone suppresses hepcidin

Testosterone increased hematocrit and hemoglobin dose-dependently, as reported (5). Six older subjects discontinued the study due to excessive erythrocytosis (hematocrit > 54%), and these subjects were not analyzed further. Hepcidin data were log-transformed to a normally distributed variable. The analysis was adjusted for the age group, its significant interactions with testosterone use, dose serum levels, and random variation within each individual. High testosterone levels resulted in a 60% suppression of serum hepcidin levels within 1 wk (87 to 35 ng/ml; P < 10⁻¹²; Fig. 1B), and all dose groups experienced at least a 50% decline in hepcidin (P < 0.0001). At this early time point, a transient testosterone “flare” occurs in response to GnRH-pituitary-testicular stimulation, with serum testosterone levels reaching supraphysiological levels (>1000 ng/dl, group data not shown) in all dose groups and then reaching equilibrium, dose-dependent levels of testosterone by 4 wk until the end of treatment (Table 2). These results support the hypothesis that testosterone suppresses hepcidin in men.

Figure 1.

A, Study design, showing simultaneous suppression of endogenous testosterone secretion (GnRH = GnRH agonist, leuprolide) and weekly injection of testosterone enanthate for 20 wk. B, Testosterone suppresses hepcidin dose-dependently in men. Log-transformed serum hepcidin levels (ng/ml) are shown at various sampling points in the 20-wk testosterone treatment study.

Table 2.

Testosterone dose-dependently suppresses serum hepcidin

Testosterone dose (mg/wk)	Midstudy testosterone (ng/ml)		Hepcidin (% of baseline)	95% CI	P value
	Younger	Older
25	135 ± 117	139 ± 79	97.2	68.4–138.0	0.8722
50	245 ± 165	263 ± 126	71.3	49.4–103.0	0.0718
125	566 ± 193	655 ± 125	47.6	33.3–67.9	<0.0001
300	1138 ± 487	1886 ± 586	34.2	23.9–49.0	<0.0001
600	2218 ± 911	3603 ± 1159	38.6	27.0–55.1	<0.0001

Multivariate analysis included testosterone dose, age, age effect/dose compared to change in serum hepcidin from baseline to midstudy (average measurement wk 4–8) in all men with comparison to testosterone dose groups shown (remaining variables are described in text). Serum testosterone levels (wk 12 by RIA) are listed alongside dose groups. CI, Confidence interval. 

To test the dose-dependency of this effect, hepcidin was measured during treatment when serum testosterone levels had reached graded, steady-state trough levels (4–20 wk). As seen in Fig. 1B, serum hepcidin levels were suppressed in the higher testosterone dose groups in younger men and dose-dependently suppressed in older men. Multivariate analysis was used to quantify the relationship between serum testosterone levels and serum hepcidin suppression. As shown in Table 2, graded increases in serum testosterone were associated with graded suppression of serum hepcidin to a maximum of 60% in the 300- and 600-mg/wk groups. Testosterone dose was highly correlated with suppression of hepcidin, and this effect remained highly significant throughout the treatment duration (Fig. 1A; P < 0.0001). We estimate that a 100-ng/dl increase in serum testosterone level was associated with a 14.9% decrease in serum hepcidin. Four weeks after discontinuation of treatment, when endogenous testosterone levels were still suppressed, serum hepcidin levels had either returned to baseline values or were higher than baseline (87 ± 70 ng/ml at baseline vs. 134 ± 107 ng/ml at wk 24; P < 0.01 for all groups). Men in the higher dose groups showed a more pronounced rebound after discontinuation of testosterone (data not shown).

Age-related effect of testosterone on hepcidin

Compared with young men, older men had higher baseline hepcidin levels (Table 1) and experienced significantly greater (1.8-fold) suppression of hepcidin levels overall (P = 0.0011). Moreover, for any dose of testosterone, older men had relatively greater suppression of hepcidin (P < 0.05) than young men. It has been previously reported that older men achieve statistically higher levels of serum testosterone in the various dose groups (10). For both young and older men, the absolute effect of testosterone on serum hepcidin levels reached a plateau at the 300-mg dose, without further suppression in the 600-mg dose group. We assessed whether serum hepcidin suppression would precede hematological changes. Indeed, hepcidin levels at 4 wk were significantly, inversely correlated with eventual hematocrit by wk 16 in older men (P = 0.001) and showed this trend in younger men (P = 0.061). Linear mixed-effects models over the course of treatment (4–20 wk) demonstrated statistically significant associations between hepcidin levels and peak hematocrit. In models controlling for baseline hematocrit, baseline hepcidin values, age (older vs. young), and study wave (wk 4, 8, 12, 16 with no assumption of linear trends in time) showed that a one log-unit change in hepcidin was associated with 0.87% mean increase in hematocrit (P < 0.001). This is consistent with a difference of approximately 0.7% hematocrit between subjects with hepcidin concentrations of 10 vs.20 ng/ml, similar to the overall trend in the combined older/young cohorts. Quartile analysis showed that men in the first quartile for change in hepcidin had significantly greater increase in hemoglobin/hematocrit (+13.7 to 14.7 g/liter/+3.5 to 4.2%) later in the study (12–16 wk) than men in the two lowest quartiles (hemoglobin/hematocrit, +1.4 to 1.8 g/liter/−0.2 to 0.2%). Thus, the men in the highest quartile of hepcidin change were at greatest risk of developing erythrocytosis during testosterone therapy. The greater effect of testosterone to increase hematocrit in older men supports the hypothesis that older men respond to testosterone with a relatively greater level of hepcidin suppression than younger men.

Correlates of hepcidin suppression

To investigate the correlates of hepcidin suppression, multivariate analyses were performed. Baseline hepcidin levels did not correlate significantly with measures of body composition, IL-6, CRP, ferritin, or testosterone levels. These variables were then analyzed after 16 wk of testosterone treatment. Serum ferritin was positively correlated with hepcidin change at the same time point (P < 0.05), whereas serum iron, iron binding capacity, and transferrin saturation did not change in response to testosterone over the course of treatment (data not shown), nor were these indices correlated with hepcidin levels (12,13). Change in lean (muscle) mass was significantly correlated with change in hepcidin in both young (P < 0.05) and older (P < 0.0001) men after treatment. Change in fat mass and the inflammatory cytokines IL-6 and CRP were not significantly correlated with change in hepcidin.

Discussion

Testosterone potently suppressed hepcidin in a dose- and age-dependent manner. Supraphysiological levels of testosterone suppressed hepcidin by more than 50%, and the suppression showed general dose dependence for 20 wk. Early changes in hepcidin levels were predictive of subsequent changes in hemoglobin and hematocrit. This study has several strengths and some limitations. The use of graded doses of testosterone resulted in a wide range of testosterone concentrations, which made it possible to analyze dose-dependent changes in hepcidin in both younger and older men. The 20-wk treatment duration was sufficiently long to adequately study the effects of testosterone on erythropoiesis. All samples were obtained in the morning, 7 d after testosterone enanthate injection. A consistent dose-response relationship between testosterone and serum hepcidin was not observed in younger men in 25-, 50-, and 125- mg/wk dosing groups, however, suggesting either that the effect of testosterone on hepcidin is different in younger men, or that this study lacked sufficient statistical power. The early GnRH-induced testosterone flare, with marked suppression of serum hepcidin, also led to a delay in our ability to examine dose-response relationships between testosterone and hepcidin. If increased iron bioavailability occurs during testosterone-induced hepcidin suppression and erythropoiesis, this process cannot be confirmed due to the lack of evidence that most iron parameters (except ferritin) remain unchanged and the lack of direct measures of erythropoiesis (reticulocytes).

Erythrocytosis is the most frequent serious adverse event associated with testosterone therapy; we propose that suppression of hepcidin contributes to this effect. Most men respond to testosterone by increasing hematocrit and hemoglobin, and the excessive erythrocytosis observed in a small number of older men may constitute a responder population. Suppressed hepcidin, in association with stable erythropoietin levels, results in increased hematocrit, suggesting an increased erythropoietic response. The mechanisms by which testosterone suppresses hepcidin levels remain unknown. Suppression of inflammation, alterations in iron-sensing, hypoxia, stimulation of erythropoiesis or direct suppression of hepcidin transcription offer candidate pathways for future study. Alterations in iron metabolism alone can result in polycythemia (15). Whether regulation of serum hepcidin levels by testosterone also affects other iron-requiring processes, such as myoglobin synthesis, iron-requiring enzymes (aconitase), or oxidative phosphorylation (cytochromes) in mitochondria remains to be investigated (16).

Age-related differences in erythropoietic response to testosterone may be related to greater suppression of hepcidin in older eugonadal men compared with younger men. Increased erythropoiesis in older men, if this accounts for the findings reported here, runs counter to prior studies showing lower erythroid proliferation in bone marrow samples from older subjects (17). Numerous mechanisms, in addition to suppressed hepcidin, could explain the greater erythropoietic response of older men to testosterone. These include age-related differences in red cell life span, restricted vascular space during erythropoiesis, differential volume response to testosterone, or as yet unknown mechanisms. Noting that unexplained anemia is correlated with elevated hepcidin levels in older men and increased clinical use of testosterone, we propose that the biological consequences of testosterone therapy on hepcidin, iron metabolism, and erythropoiesis should be investigated further (18,19).