Dapagliflozin Suppresses Hepcidin And Increases Erythropoiesis

Abstract

Context

Dapagliflozin and other SGLT2 inhibitors are known to increase hematocrit, possibly due to its diuretic effects and hemoconcentration.

Objective

Since type 2 diabetes is a proinflammatory state and since hepcidin, a known suppressor of erythropoiesis, is increased in proinflammatory states, we investigated the possibility that dapagliflozin suppresses hepcidin concentrations and thus increases erythropoiesis.

Design

Prospective, randomized, and placebo-controlled study.

Setting

Single endocrinology center.

Patients

Fifty-two obese type 2 diabetes patients.

Intervention

Patients were randomized (1:1) to either dapagliflozin (10 mg daily) or placebo for 12 weeks. Blood samples were collected before and after treatments and serum, plasma, and mononuclear cells (MNC) were prepared.

Main Outcome Measure

Hepcidin and other hematopoietic factors.

Results

Following dapagliflozin treatment, there was a significant fall in HbA1c and a significant increase in hemoglobin concentration and hematocrit. Dapagliflozin treatment significantly reduced circulating hepcidin and ferritin concentrations while causing a significant increase in levels of the hepcidin inhibitor, erythroferrone, and a transient increase in erythropoietin. Additionally, dapagliflozin increased plasma transferrin levels and expression of transferrin receptors 1 and 2 in MNC, while there was no change in the expression of the iron cellular transporter, ferroportin. Dapagliflozin treatment also caused a decrease in hypoxia-induced factor-1α expression in MNC while it increased the expression of its inhibitor, prolyl hydroxylase-2. There were no significant changes in any of these indices in the placebo group.

Conclusions

We conclude that dapagliflozin increases erythropoiesis and hematocrit through mechanisms that involve the suppression of hepcidin and the modulation of other iron regulatory proteins.

iron, hepcidin, dapagliflozin, erythropoiesis

The use of sodium glucose co-transporter 2 (SGLT2) inhibitors are associated with an increase in hematocrit, of which the mechanism underlying it is not known (1, 2). The initial contraction in plasma volume following the glycosuria induced by these drugs is transient and, therefore, it does not explain the persistent increase in hematocrit (3). Also, in a comparative study of dapagliflozin and hydrochlorothiazide (a diuretic), a reduction in plasma volume and an increase in erythrocyte mass was noted with dapagliflozin but not with hydrochlorothiazide (over a 12-week period of treatment) (4). It has been suggested that an increase in hematocrit may, through an increase in oxygen-carrying capacity, contribute to the cardioprotective effect of this class of drugs (5, 6). It is, therefore, important to investigate the underlying mechanisms of this increase in hematocrit.

Hepcidin, a protein produced mainly in the liver, binds to ferroportin in the plasma membrane and thus inhibits its ability to transport iron. Hepcidin is known to be increased in anemia of chronic inflammation (7). The binding of hepcidin to ferroportin leads to tyrosine phosphorylation, ubiquitination, and the eventual proteolytic degradation of the latter (7). Hepcidin has recently been shown to also directly inhibit iron transport by ferroportin by binding to it and preventing the attachment of iron to ferroportin (8). Since ferroportin is responsible for the absorption of iron from the intestinal cell and also its export from the macrophage into the circulation, the inhibition and suppression of ferroportin is likely to reduce the bioavailability of iron. On the other hand, erythroferrone, a protein produced by erythroblasts, inhibits the expression of hepcidin, and so increases the amount of iron available for hemoglobin synthesis (9). Erythropoiesis is also activated through the hypoxia inducible factor1-α (HIF1-α), which stimulates erythropoietin (EPO) formation. At physiological normoxic conditions, prolyl hydroxylase domain-containing proteins (PHDs) hydroxylate proline residues on HIF1-α subunits, leading to their destabilization, ubiquitination, and subsequent proteasomal degradation. In hypoxic conditions, the O₂-dependent hydroxylation of HIF1-α subunits by PHDs is reduced, resulting in HIF1-α accumulation, dimerization with HIF1-β, and migration into the nucleus to induce an adaptive transcriptional response, including increasing expression levels of EPO.(10).

We have recently shown that the increase in hematocrit following testosterone treatment in male patients with hypogonadism is not only associated with an increase in EPO but also with the suppression of hepcidin and an increase in the expression of ferroportin and an increase in transferrin receptors and transferrin concentrations (11). Thus, the reversal of anemia in male hypogonadism following testosterone treatment is due to multiple mechanisms. These mechanisms may be relevant to the action of SGLT2 inhibitors.

Based on the above, we hypothesized that dapagliflozin, an SGLT2 inhibitor, suppresses hepcidin while increasing plasma concentrations of EPO, erythroferrone, transferrin, and the expression of ferroportin and transferrin receptors.

Patients and Methods

Patients

This is a single-center, prospective, randomized double-blind, and placebo-controlled study conducted at the Endocrinology, Diabetes and Metabolism, research center at the University at Buffalo. Fifty-two patients with type 2 diabetes (HbA1c < 8) and obesity on oral antidiabetic agents were randomized (1:1) to either dapagliflozin or placebo for 12 weeks. There were 5 drop-outs from the study (3 in the placebo group and 2 in the dapagliflozin group); 24 patients in the dapagliflozin group and 23 patients in the placebo group completed all study visits and have evaluable data. The drop-outs were due to personal reasons, with no safety or protocol-related concerns. Glucagon-like peptide-1 receptor agonists (GLP-1 RAs) and SGLT2 inhibitor use was exclusionary. Patients on statins, angiotensin-converting enzyme (ACE) inhibitors, angiotensin receptor blockers (ARBs), diuretics, dipeptidyl peptidase 4 inhibitors, thiazolidinediones, and antioxidants were allowed as long as they were on stable doses of these compounds and the dosage did not changed during the course of the study. Patients were mostly on metformin (85%), insulin (38%), sulfonylureas (38%), and pioglitazone (21%) for their diabetes management. In addition, about 65% of patients were on statins, 47% on ACE inhibitors, 38% on ARBs, and about 28% on diuretics or beta blockers. There was no significant difference between the dapagliflozin and placebo groups in proportion to the patients on any one of these other medications. Patients were started on a 5 mg daily dose, which was titrated to 10 mg daily during the first week. Fasting blood samples were collected at baseline and at 1, 6, and 12 weeks of treatment. All patients provided informed consent and signed a consent form. The study was approved by the University at Buffalo institutional review board and was registered in Clinicaltrial.gov under registration number: NCT02433678. Patients’ baseline characteristics were not significantly different between the 2 groups and are summarized in Table 1.

Table 1.

Baseline and post-treatment changes in demographic data.

	Baseline	At 12 Weeks	Baseline	At 12 Weeks	Change in Dapagliflozin vs. Change in Placebo (95% CI)	P
	Placebo (n = 23)		Dapagliflozin (n = 24)
Age (years)	62 ± 2		61 ± 2
BMI (kg/m2)	36.9 ± 2.1	37.0 ± 2.1	39.1 ± 1.8	38.7 ± 1.8	-0.5 (-1.2 to 0.2)	0.13
Weight (Kg)	110.6 ± 6.2	111.1 ± 6.3	112.4 ± 5.3	111.4 ± 5.2	-1.5 (-3.5 to 0.5)	0.14
HbA1c	6.7 ± 0.2	6.8 ± 0.2	6.9 ± 0.1	6.6 ± 0.1*	-0.4 (-0.7 to -0.1)	0.01
Fasting glucose (mg/dl)	145 ± 12	152 ± 19	142 ± 13	125 ± 11*	-24 (-44 to -4)	0.014
Fasting insulin (μU/ml)	15.3 ± 4.2	14.2 ± 4.1	16.5 ± 3.3	20.1 ± 5.6	4.7 (-0.8 to 10.2)	0.14
Systolic blood pressure	136 ± 3	133 ± 4	134 ± 3	124 ± 2*	-7 (-12 to -2)	0.021
Diastolic blood pressure	80 ± 2	74 ± 4	78 ± 2	77 ± 2	5 (-4 to 14)	0.43
Hemoglobin (g/dl)	13.1 ± 0.4	13.1 ± 0.4	13.4 ± 0.3	13.9 ± 0.4*	0.5 (0.2 to 0.8)	0.020
Hematocrit (%)	40.9 ± 1.0	41.0 ± 0.8	41.3 ± 0.9	43.5 ± 1.0*	2.1 (0.9 to 3.3)	0.021
RBC (mil/mm3)	4.7 ± 0.1	4.7 ± 0.1	4.7 ± 0.1	4.9 ± 0.1*	0.2 (0.1 to 0.3)	0.011
MCV (fL)	87 ± 1	87 ± 1	89 ± 1	88 ± 1	-1 (-2.2 to 0.2)	0.13
Iron (µg/dL)	92 ± 6	90 ± 5	92 ± 5	80 ± 6*	-10 (-23 to 3)	0.10
Transferrin (mg/dL)	296 ± 11	291 ± 10	298 ± 10	315 ± 11*	22 (14 to 30)	0.027
Transferrin saturation (%)	24 ± 2	24 ± 2	24 ± 1	20 ± 2*	-4 (-6 to -2)	0.005
Total IBC (µg/dL)	380 ± 12	382 ± 12	373 ± 12	395 ± 13*	20 (-6 to 46)	0.097
Unsaturated IBC (µg/dL)	283 ± 13	281 ± 12	282 ± 14	323 ± 15*	43 (18 to 68)	0.018
Serum creatinine (mg/dl)	0.99 ± 0.04	0.97 ± 0.04	0.89 ± 0.04	0.96 ± 0.05*	0.09 (0.04 to 0.14)	0.007
eGFR (mL/min/1.73m2)	81 ± 5	82 ± 5	89 ± 5	85 ± 5*	-5 (-8 to -3)	0.008
Urine volume/24hr (L)	1.85 ± 0.19	1.80 ± 0.19	1.91 ± 0.15	1.92 ± 0.17	0.06 (-0.02 to 0.14)	0.12
Aspartate aminotransferase (U/L)	25 ± 3	24 ± 2	28 ± 3	22 ± 1*	-5 (-9 to -1)	0.029
Alanine aminotransferase (U/L)	24 ± 4	23 ± 2	27 ± 2	24 ± 2	-2 (-5 to 1)	0.19

Data is presented as mean ± SE. Statistical comparisons of the change from baselines are made within groups using paired t-test (*P < 0.05). P values represent analysis using t-test for the difference in the changes between the groups.

Abbreviations: CI, confidence interval; IBC, iron binding capacity.

MNC isolation.

Blood samples were collected in K₃-EDTA and layered on lympholyte medium (Cedarlane Laboratories, Hornby, Ontario, Canada). The mononuclear cell (MNC) band was harvested and washed twice with Hank’s balanced salt solution.

Quantification of mRNA expression by RT-PCR.

Expression of iron regulatory mediators were measured in MNC by RT-PCR. Total RNA was isolated from MNC using commercially available RNAqueous^®-4PCR Kit (Ambion, Austin, Texas). Real-time RT-PCR was performed using Stratagene Mx3000P QPCR System (La Jolla, California), Sybergreen master mix(Qiagen, California), and gene-specific primers for ferroportin, ferritin, transferrin receptors (TR) 1 and 2, HIF1α, and PHD2 (Life Technologies, Carlsbad, CA). All values were normalized to the expression of a group of housekeeping genes, including actin, ubiquitin C, and cyclophilin A. The normalization factor used is calculated by geNorm software and is based on the values of all housekeeping genes used.

Plasma measurements.

Complete blood counts, HbA1c, plasma concentrations of iron, transferrin, transferrin saturation, total iron binding capacity (TIBC), and unsaturated iron binding capacity (UIBC) were calculated or measured by well-established clinical laboratory assays. Enzyme-linked immunosorbent assay (ELISA) was used to measure concentrations of hepcidin (intra- and interassay CVs are 8% and 12%, respectively), erythroferrone (MyBiosource, San Diego, California), and ferritin (Elabscience, Houston, Texas).

Statistical analysis.

This is a post hoc analysis of samples collected in a study investigating the anti-inflammatory effects of dapagliflozin. Additional testing was conducted using SigmaStat software (SPSS Inc., Chicago, Illinois). All data are represented as mean ± standard error (SE). Baseline measurements were also normalized to 100% and changes from baseline were calculated. Statistical analysis of the change in hepcidin was carried out using analysis of variance (ANOVA) for changes compared to baseline and 2-factor ANOVA Dunnett’s post hoc test for multiple comparisons between different treatments within each 2-factor ANOVA test. Similar tests were used for secondary endpoints and were analyzed using similar tests or the paired t-test compared to baseline or the t-test for group comparisons. Adjustment for multiple testing was not performed for this post hoc analysis study.

Results

Metabolic changes

In the dapagliflozin group, there was a significant fall in HbA1c from 6.9 ± 0.1% to 6.6 ± 0.1% (P = 0.01 compared to placebo), fasting blood glucose by 17 mg/dl (from 142 ± 13 to 125 ± 11 mg/dl [P = 0.014 compared to placebo]), and in systolic blood pressure by 10mmHg (from 134 ± 3 to 124 ± 2 mmHg [P = 0.021]), while there was an increase in creatinine levels and a decrease in aspartate aminotransferase (AST) (Table 1). There was no significant change in body weight or fasting insulin levels in either group.

Hemoglobin concentrations and hematocrit and red blood cell parameters

In the dapagliflozin group, there was an increase in hemoglobin concentration from 13.4 ± 0.3 to 13.9 ± 0.4 g/L (P = 0.02) and in hematocrit from 41.3 ± 0.9% to 43.5 ± 1.0% (P = 0.02). There was also a significant increase in red blood cell count and a modest decrease in mean corpuscular volume (Table 1).

Iron regulatory mediators in plasma

Hepcidin decreased by 24 ± 3% (from 193 ± 24 to 149 ± 20 ng/ml, placebo-corrected change of -46 ± 6 ng/ml, P = 0.007 [Fig. 1A]) following 12 weeks of dapagliflozin treatment, while erythropoietin increased transiently at 6 weeks of treatment by 21 ± 7% (from 11.3 ± 1.4 to 17.4 ± 2.5 mU/ml). This increase was significantly greater than the baseline, but the placebo-corrected change of 5.8 ± 0.8 mU/ml, P = 0.074 was short of significance (Fig. 1B). There was also a significant fall in ferritin concentrations by 32 ± 7% (from 10.75 ± 1.32 to 5.92 ± 0.88 ng/ml, placebo-corrected change of -4.23 ± 0.63 ng/ml P = 0.011 [Fig. 1]), while transferrin concentration increased by 11 ± 3% (from 298 ± 10 to 315 ± 11 mg/dl, placebo-corrected change = 22 ± 4 mg/dl, P = 0.027 [Fig. 1D]). In addition, there was an increase in levels of the hepcidin inhibitor, erythroferrone, by 71 ± 22% (from 1.48 ± 0.73 to 2.98 ± 1.61 ng/ml, placebo-corrected change of 1.47 ± 0.63 ng/ml P = 0.024 [Fig. 1E]), while serum iron concentration decreased transiently at 6 weeks by 16 ± 6% after dapagliflozin therapy as compared to placebo (P = 0.035 [Fig. 1F]).

Figure 1.

Plasma levels of hepcidin (A), EPO (B), ferritin (C), erythroferrone (D), transferrin (E), and iron (F) before and following 12 weeks of treatment with placebo or 10 mg daily of dapagliflozin in type 2 diabetes patients. Data presented as mean ± SE, * = P < 0.05 compared to baseline in dapagliflozin group using ANOVA and paired t-test and #=P < 0.05 compared to placebo using 2-way ANOVA and t-test.

Consistent with this, transferrin saturation decreased by 22 ± 5% (P = 0.005), while total iron binding capacity increased by 10 ± 3% (P = 0.025 vs. baseline) and unsaturated iron binding capacity increased by 18 ± 3% (P = 0.013 [Table 1]).

Baseline hepcidin levels and changes in hepcidin levels were not related to baseline levels or changes in hemoglobin or to erythropoietin.

Expression of iron transport mediators in MNC

Both TR1 and TR2 mRNA expression in MNC increased significantly by 59 ± 14% and 62 ± 14%, respectively, after 12 weeks of dapagliflozin treatment (P < 0.01 [Fig. 2A and 2B]) compared to baseline. However, only the increase in TR1 expression was significantly different compared to the placebo group. Ferritin mRNA expression decreased significantly by 46 ± 5% (P < 0.001 compared to placebo [Fig. 2C]), while there was no change in ferroportin expression (Fig. 2D) following dapagliflozin treatment. There was no change in the expression of any these indices in the placebo group.

Figure 2.

mRNA expression of transferrin receptor 1 and 2 (A–B), ferritin (C), ferroportin (D), hypoxia-induced factor (HIF)1α (E), and prolyl hydroxylase domain-containing protein-2 (PHD2) (F) in MNC before and following 12 weeks of treatment with placebo or 10 mg daily of dapagliflozin in type 2 diabetes patients. Data presented as mean ± SE, * = P < 0.05 compared to baseline in dapagliflozin group using ANOVA and paired t-test and # = P < 0.05 compared to placebo using 2-way ANOVA and t-test.

HIF1α and prolyl hydroxylase domain-containing protein-2 (PHD2) expression

HIF1α mRNA expression in MNC decreased significantly by 42 ± 15% starting at week 6 (P = 0.015 compared to placebo) and by 32 ± 12% at week 12 following dapagliflozin treatment (P = 0.022 compared to baseline [Fig. 2E]). On the other hand, there was a significant increase (by 38 ± 13% at week 12, P = 0.017 compared to placebo [Fig. 2F]) in the expression of PHD2 following dapagliflozin treatment.

Discussion

Our data clearly show that dapagliflozin administration for 12 weeks in patients with type 2 diabetes results in an increase in hematocrit, the suppression of hepcidin, and an increase in plasma concentrations of transferrin and the expression of transferrin receptors 1 and 2. In addition, there was a fall in plasma ferritin and iron concentrations. The fall in ferritin and iron concentrations indicates a greater mobilization and utilization of stored iron. These processes enable the release of iron from intracellular stores and its transport from the stored sites to utilization sites, where it is taken up by the transferrin receptor for the synthesis of iron-containing proteins, including hemoglobin. There was a transient increase in erythropoietin at 6 weeks that was significantly greater from the baseline but not significantly different from the placebo group. There was also a significant increase in plasma concentrations of erythroferrone, which is an inhibitor of hepcidin. Thus, there is a dual inhibitory effect on hepcidin: a reduction in its concentration and an increase in its inhibitor.

However, there was no change in ferroportin expression. Since hepcidin suppresses the release of iron from storage sites by suppressing ferroportin expression, and since the expression of ferroportin did not alter following dapagliflozin administration, it is possible that the actual inhibition of iron transport through ferroportin by hepcidin may be responsible. It has recently been shown that hepcidin binds to ferroportin and prevents the binding of iron to the intramolecular channel through which iron gets transported (8, 12). Transferrin concentrations and transferrin receptor expression, both of which increased after dapagliflozin, may also have been affected by hepcidin. The fall in plasma ferritin concentration is consistent with greater peripheral tissue uptake of iron from plasma.

The other important observation was that HIF1α expression was suppressed. Since HIF1α promotes erythropoiesis, its reduction was unexpected. However, it is possible that this mechanism is not involved in mediating the effect of dapagliflozin and that the increase in hematocrit and the consequent increase in oxygen-carrying capacity by other mechanisms described above suppresses HIF1α expression. It is also of interest that proline hydroxylase-2 expression also increased, which is consistent with the reduction in HIF1α expression.

The sequence of action on iron regulatory proteins after dapagliflozin was similar to that after the administration of testosterone in patients with hypogonadotropic hypogonadism (HH) and type 2 diabetes,as previously described. Testosterone also induced an increase in hematocrit, erythropoietin, transferrin, and transferrin receptors over a period of 15 weeks. However, testosterone induced a marked increase in ferroportin expression, which was not observed with dapagliflozin. Whether an increase in ferroportin expression is involved in an increase in hematocrit, as observed in longer-term dapagliflozin studies, requires further investigation. It would appear that during the chronic inflammatory conditions of type 2 diabetes, as with HH, the increase in hepcidin leads to the suppression of the release of stored iron. This process is reversed following appropriate treatment.

It is of interest that combined testosterone and SGLT2 inhibitor treatment in type 2 diabetes patients may cause erythrocytosis (13). Therefore, and in view of recent label expansion and off-label use of some SGLT2 inhibitors in conditions like type 2 diabetes with heart failure and in type 1 diabetes, caution should be observed when combined testosterone and SGLT2 inhibitors use is considered.

While the exact mechanisms by which dapagliflozin suppresses hepcidin is not fully known, the data from this study and other studies support an early role for EPO which is produced by the kidneys following dapagliflozin treatment (14). EPO upregulates erythroferrone production which is a known suppressor of hepcidin transcription in the liver (15) (Fig. 3). In addition, SGLT2 inhibitors, through glycosuria, increased glucagon secretion and increased β-oxidation of fatty acids in the liver, and they are known to suppress de novo lipid synthesis in the liver (16). This can potentially contribute to the suppression of hepatic fat accumulation and inflammation and contribute to the improvement of liver function, especially in conditions such as nonalcoholic steatohepatitis (NASH) and fatty liver disease. This is supported by recent studies in which empagliflozin and dapagliflozin have been shown to reduce hepatic fat and other markers of liver fibrosis (17, 18). While we did not investigate hepatic inflammation or fat accumulation in our study, dapagliflozin reduced the liver enzymes AST and alanine aminotransferase (ALT) (Table 1), an indicator of improved liver function. Since hepcidin and ferritin are produced mainly by the liver and are associated with proinflammatory states—and since dapagliflozin has a significant and rapid suppressive effect on both of them—it is possible that the combined anti-inflammatory and lipolytic effect of dapagliflozin on the liver can contribute to hepcidin suppression as part of the overall improvement in liver function (Fig. 3).

Figure 3.

Schematic diagram showing possible mechanisms of dapagliflozin suppression of hepcidin. Dapagliflozin induces an early increase in erythropoietin (EPO) secretion from the kidneys, which upregulates erythroferrone, a hormone that suppresses hepcidin expression in the liver. Additionally, dapagliflozin induces glycosuria and increases glucagon secretion from α-cells in the pancreas. These changes suppress lipid synthesis and increase β-oxidation, which leads to reduced hepatic fat and inflammation. Suppression of chronic inflammation associated with fatty liver can play a role in hepcidin suppression. It is also possible that dapagliflozin affects the liver directly in mechanisms that are currently unknown (?).

In conclusion, dapagliflozin induced a mild but significant increase in hematocrit, which is consistent with previously published data. There was, in parallel, a significant decrease in plasma hepcidin concentration and an increase in plasma transferrin concentration and the expression of transferrin receptors. A transient increase in EPO concentration was also observed. There was a concomitant fall in plasma concentrations of ferritin. These data are consistent with the release and utilization of iron, both of which are known to be inhibited in chronic inflammatory states. The increase in hematocrit and oxygen-carrying capacity probably induced an increase in hydroxylase activity and suppressed HIF1α expression. Thus, while the mechanisms underlying the increase in hematocrit following dapagliflozin administration are complex, hepcidin suppression-related increase in bioavailability and the transport and utilization of iron in hematopoiesis appear to be the major factors contributing to the increase the hematocrit following treatment with dapagliflozin.

Acknowledgments

Clinical Trial Information: Clinicaltrial.gov registration no. NCT02433678

Additional Information

Disclosure Summary: The original clinical trial is supported by a research grant from AstraZeneca Pharmaceuticals LP (Grant 10400) to Paresh Dandona, who is also a consultant and a speaker for AstraZeneca Pharmaceuticals LP. The other authors have nothing to disclose.

Data availability: The datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.