Serum hepcidin level, iron metabolism and osteoporosis in patients with rheumatoid arthritis

Hepcidin, a major regulator of iron metabolism and homeostasis, is regulated by inflammation. Recent studies have suggested that hepcidin and iron metabolism are involved in osteoporosis, and the aim of this study was to determine whether serum hepcidin levels are correlated with the degree of osteoporosis in patients with rheumatoid arthritis (RA). A total of 262 patients with RA (67.5 ± 11.4 years; 77.5% female) were enrolled. Serum iron, ferritin, and hepcidin levels were positively correlated each other. Multiple regression analyses revealed that the serum iron level was positively correlated with femoral T and Z scores, whereas the serum hepcidin level was not. Serum hepcidin level was correlated with the serum 25-hydroxy vitamin D level, which was in turn positively related to the femoral Z score. Serum hepcidin and serum iron were indirectly and directly related to osteoporosis in patients with RA.


Associations among the serum hepcidin level, iron metabolism, and inflammation. Serum iron,
ferritin, and hepcidin levels were significantly positively correlated ( Table 2). The Hb level was slightly positively related to these markers of iron metabolism, and platelet count was negatively related to these markers. Markers of inflammation due to RA (DAS28-CRP and serum CRP) were negatively associated with the serum iron level and positively related to serum ferritin and hepcidin levels. With regard to categorical markers, the serum iron level was significantly lower in patients who used PSL than those who did not (83.1 ± 31.4 vs. 69.5 ± 34.9 mg/ dL, p = 0.001) ( Table 3). Lower serum iron and higher serum ferritin and hepcidin levels were seen in patients who used bDMARDs including TCZ than those who did not (iron, 69.8 ± 32.8 vs. 79.4 ± 36.9 mg/dL, p = 0.046; ferritin, 68.7 ± 69.2 vs. 50.1 ± 48.9 ng/mL, p = 0.023; hepcidin, 15.8 ± 20.2 vs. 11.3 ± 18.4 ng/mL, p = 0.008), and only higher serum iron level was seen in patients who used TCZ (71.2 ± 33.8 vs. 93.4 ± 36.9 mg/dL, p = 0.007).
Multiple regression analyses adjusted for age, sex, BMI, eGFR, CRP, the use of bDMARD, PSL, anti-bone resorption drug, teriparatide, and iron agent, indicated that the serum iron level was positively related to femoral T and Z scores (β = 0.121, p = 0.039 and β = 0.123, p = 0.046, respectively), and serum ferritin and hepcidin levels were positively related to the serum 25(OH)D level (β = 0.200, p = 0.002 and β = 0.207, p = 0.002, respectively). Annual change of lumbar T score, but not that of femoral T score, was positively related to serum hepcidin level (β = 0.159, p = 0.025). MMP-3 was also positively related to the serum ferritin level (β = 0.160, p = 0.004), whereas FGF23, bone alkaline phosphatase and TRACP-5b were not related to any of these three variables (Table 4).
After stratification according to disease activity, a positive relationship between serum iron and the femoral T score was observed (r = 0.318, p = 0.002), and between serum hepcidin and the annual change in lumbar T score (r = 0.293, p = 0.010) in patients in remission (CDAI ≦2.8, n = 93) (Table S2). Serum 25(OH)D level was positively associated with serum ferritin and serum hepcidin levels in patients in remission (r = 0.356, p < 0.001 and r = 0.342, p = 0.001; respectively, n = 93) and in those showing low disease activity (r = 0.293, p = 0.001 and r = 0.204, p = 0.023, respectively; n = 123) but not in those showing moderate to high disease activity (Table S2).
Associations between serum hepcidin level and other parameters. All subjects were divided into four groups according to serum hepcidin level (Q1, -1.3 ng/mL; Q2, 1.4-6.6 ng/mL; Q3, 6.7-19.5 ng/mL; Q4, 20.0-102.3 ng/mL; Table S3). Higher hepcidin levels were significantly related to greater inflammation, a higher ferritin level, a higher iron level, lower unsaturated iron binding capacity (UIBC), higher 25(OH)D level, and  Fig. 3). Lower hepcidin levels were significantly related to a higher daily PSL dose, a higher rate of bDMARD use, a higher platelet count, and a lower Hb level. Age, disease duration, renal function, TCZ use, BMD, markers of bone metabolism, and FGF23 level were not related to the serum hepcidin level (Table S3, Fig. 3). Iron metabolism, inflammation due to RA, and the 25(OH)D level were significantly related to the serum hepcidin level.

Associations between the serum 25(OH)D level and other parameters.
All subjects were divided into four groups according to serum 25(OH)D level (Q1, 3.5-11.6 ng/mL; Q2, 11.7-15.6 ng/mL; Q3, 15.8-20.6 ng/ mL; Q4, 20.7-50.3 ng/mL; Table S4). Higher 25(OH)D levels were significantly related to older age, a higher adjusted Ca level, a lower eGFR level, a higher ferritin level, a higher hepcidin level, a higher Z score of femoral neck, a higher FGF23 level and a higher annual change of lumbar T score (Table S4, Fig. 3). Inflammation, disease activity, and treatments were not associated with serum 25(OH)D level. The same analyses were performed on patients who did not use active vitamin D (non-VD users) only, and similar results were obtained (Table S5).

Discussion
The results of this study indicate that the serum iron level is positively related to BMD and serum hepcidin and ferritin levels are positively related to the 25(OH)D level, which is positively related to the femoral Z score. As to 231 patients re-examined BMD, annual change of lumbar T score was positively related to serum hepcidine level and higher 25(OH)D quartiles. The serum hepcidin level and iron metabolism indirectly and directly affected osteoporosis in patients with RA.
Researchers have reported serum hepcidin concentrations in patients with RA, focusing on anemia. In studies, the serum hepcidin level was higher in patients with RA compared to healthy subjects 15,16 , which may be the result of inflammation due to RA. Iron deficiency, which often accompanies RA, is also an essential factor influencing serum hepcidin. In one study, patients with RA and iron deficiency had significantly decreased serum hepcidin levels compared to those with RA and anemia with chronic inflammation 17 . As hepcidin is influenced by inflammation and iron metabolism, results differ on the influence on Hb and disease activity associated with hepcidin on RA 16,18 . Furthermore, in the same study of patients with RA, iron metabolism was related to serum hepcidin levels cross-sectionally, but inflammation was related longitudinally 19 . Anti-IL-6 therapy decreases serum hepcidin and improves the Hb level [20][21][22] , and the effect is more marked with anti-IL-6 than anti-TNF therapy 20 . In our cross-sectional study, CRP and Hb were related to the serum hepcidin level, but stronger relationships were Mean ± S.D.

Min -Max
Lumbar Z score 0 www.nature.com/scientificreports www.nature.com/scientificreports/ observed for iron metabolism (i.e., ferritin, iron, and UIBC). Moreover, positive relationships were found among serum hepcidin, ferritin, iron and Hb levels which seemed like iron deficiency anemia pattern. The reason for less influence from inflammation was considered because most patients in this study mainly showed low disease activity or were in remission. Also a high percentage of patients in the lower quartile of serum hepcidin level were treated with bDMARDs. Thus, the serum hepcidin level seemed to be closely reflected by iron metabolism, including in patients showing low disease activity or in remission.
Iron accumulation is a risk factor for osteoporosis, and hepcidin is expected to be a useful therapeutic target [23][24][25] . In one study, hepcidin knockout mice had a higher serum ferritin level and higher iron in the liver and femur than controls and showed low bone mass and changes in bone microarchitecture 26 . Hepcidin knockout mice also showed a marked reduction in bone load-bearing capacity with enhanced bone resorption 23 . A mouse model with overexpression of hepcidin showed higher levels of serum hepcidin and lower levels of serum ferritin, and bone loss and changes in markers of bone metabolism after ovariectomy were ameliorated 24 . In humans, genetic hemochromatosis and thalassemia cause iron overload, and osteoporosis is a major complication 11,12 . Iron overload (an elevated ferritin level) is a risk factor for progressive bone loss in healthy postmenopausal women and middle-aged men and a risk factor for radiological vertebral fracture in postmenopausal women 13 . Liu et al. compared serum hepcidin levels in 40 patients with osteoporosis and 40 healthy controls 14 . They identified lower serum hepcidin levels and higher iron levels in patients with osteoporosis compared to healthy controls, and the serum hepcidin level was negatively related to the serum iron level 14 . In this study, the serum hepcidin level was positively related to serum iron and ferritin levels, and serum iron levels were positively related to BMD, in contrast to the studies outlined above. Moreover, annual change of lumbar T score, not femoral, was positively related to serum hepcidin and 25(OH)D levels. No direct relations were found between markers of bone metabolism and the serum hepcidin level. Those findings were more apparent in patients who were in remission. The unexpected results about iron metabolism and osteoporosis in this study seemed because the relationships between serum iron and hepcidin levels and inflammation due to RA, bDMARD treatment, and iron deficiency in patients with RA are complicated. The disease activity of most patients in this study was low; the results may have been different in patients with higher RA activity.
Vitamin D deficiency is a risk factor for autoimmune disorders, including RA 27 . Vitamin D affects bone mineralization and calcium regulation, and the serum level of 25(OH)D in RA is positively related to BMD 28 . A recent in vitro study suggested that binding of 1,25(OH) 2 D to the vitamin D receptor directly suppressed hepcidin gene transcription 29 . Furthermore, supplementation of vitamin D reduces serum hepcidin levels in healthy subjects 29,30 , patients with chronic kidney disease 29,31 , and pediatric patients with inflammatory bowel disease 32 . In this study, the serum 25(OH)D level was not related to disease activity but was negatively related to renal function and positively related to the serum adjusted Ca level, serum ferritin level, serum hepcidin level, serum FGF23 level, and femoral Z score. There are some possible explanations for the positive association between 25(OH)D and hepcidin. First, the results of cross-sectional studies of the relationship between hepcidin and 25(OH)D are inconsistent. One study of children with inflammatory bowel disease suggested that a higher 25(OH)D concentration was related to a lower hepcidin level 33 . Another study of older Mexican adults indicated that the serum hepcidin level did not differ between patients with 25(OH)D ≥ 50 nmol/L and <50 nmol/L 34 . The characteristics  www.nature.com/scientificreports www.nature.com/scientificreports/ of subjects were markedly different in these two studies and the present study. Second, race may influence the association between vitamin D deficiency and anemia 35 . In one study, serum 25(OH)D < 50 nmol/L was significantly associated with anemia among black but not white subjects 35 . Further research is needed to reach definitive conclusions.
In this study, a higher FGF23 level was related to less kidney function and older age but not to serum hepcidin level. As we reported previously 2 , RA disease activity and MMP-3 are positively related to serum FGF23 levels. Meanwhile, this study showed that higher FGF23 levels are related to lower Hb levels and to higher 25(OH)D and adjusted Ca levels. This study and the previous study differed with regard to the subjects; the previous study included patients with higher disease activity (CRP, 3.2 ± 3.4 mg/dL; DAS28-ESR, 4.7 ± 1.4; bDMARD use, 6.6%) than patients in the present study.
Finally, we investigated relationships among serum hepcidin, iron metabolism and osteoporosis in patients with RA (Fig. 4). Hepcidin is usually regulated by iron metabolism (iron overload leads to an increase in hepcidin, and iron deficiency leads to a decrease) and is suppressed by erythropoiesis, sex hormones, and growth factors 36 . Hepcidin deficiency due to genetic hemochromatosis or severe liver dysfunction leads to higher iron levels, while higher hepcidin levels decrease iron levels, as seen in chronic inflammation and some cancers. When inflammation occurs due to RA, the production of hepcidin increases due to the expression of inflammatory cytokines, and affects to decreasing iron and increasing ferritin levels. In this study of RA patients in remission or showing low disease activity, the serum iron level was positively associated with serum hepcidin and ferritin levels, but was negatively associated with inflammation due to RA. Although iron overload and hepcidin may influence osteoporosis, the serum iron level was positively related to BMD in this study but serum hepcidin and ferritin levels were not. However, the serum 25(OH)D level was positively related to the serum hepcidin level and also positively related to femoral Z scores. Serum hepcidin level was also positively related to the annual change of lumbar T score. In contrast to previous reports, opposite effects of the serum iron level to BMD and the serum 25(OH)D level to the serum hepcidin level were indicated, and further research is needed to determine the mechanisms. However, the serum FGF23 level was not directly related to the serum hepcidin level, but the serum 25(OH)D level and inflammation were common factors related to both serum hepcidin and FGF23 levels.
The major limitation of this study is single-center nature and the absence of healthy controls. Blood samples were not collected during fasting so an effect of diet cannot be ruled out. We could not analyze the patient background of smoking, alcohol intake, menopausal status and the use of hormone replacement therapy. Also, physical activity and body weight bearing activity were not included, but health assessment questionnaire without disability index (HAQ-DI) and BMI could replace them. We investigated only serum markers and BMD but not fracture data. Further studies are needed regarding to fragility fractures controlling patients' background. The number of patients using iron agents was low and no significant effect was observed.
In conclusion, the serum iron level was positively related to BMD in these patients with RA, and a higher iron level was not considered a risk factor for osteoporosis. The serum hepcidin level was not related to BMD and markers of bone metabolism but was positively related to the serum 25(OH)D level, which was positively related  Bone mineral density measurement. Bone mineral density (BMD) using dual-energy X-ray absorptiometry (PRODIGY; GE Healthcare, Madison, WI, USA). Lumbar spine L1-4 and basically left, but if impossible right femoral neck, were evaluated at the recruitment in all patients. The T score represents the difference from the mean BMD of young healthy subjects (in terms of standard deviation), and the Z score is obtained based on comparison with age-and sex-matched controls. The BMD T scores were classified as follows: ≦−2.5, osteoporosis; −2.5< and < −1, osteopenia; −1≦, normal. Two hundred and thirty-one patients were re-examined BMD in about three years and annual change of T score was evaluated. The mean observational period was 2.0 ± 0.33 years (0.63-3.25).   Table 4. Relationships between serum iron, ferritin, and hepcidin levels and osteoporosis-related factors analyzed by multiple regression. Multiple regression models were adjusted for age, sex, BMI, eGFR, CRP, PSL use, bDMARD use, anti-resorption drug use, teriparatide use, and iron agent use. TRACP-5b, tartrate-resistant acid phosphatase-5b; 25(OH)D, 25-hydroxy vitamin D; FGF23, fibroblast growth factor 23; MMP-3, matrix metalloprotease 3. *Annual change of T score was evaluated in 231 patients and the mean observational period was 2.0 ± 0.33 years (0.63-3.25). (2020) 10:9882 | https://doi.org/10.1038/s41598-020-66945-3 www.nature.com/scientificreports www.nature.com/scientificreports/ Statistical analyses. Most data are expressed as the mean ± standard deviation or number (%). As serum hepcidin levels were not normally distributed, the actual hepcidin level + 1 was logarithmically transformed. Correlation coefficients were obtained using Spearman's rank method and also correlation coefficients adjusted  . Estimated relationships related to hepcidin in patients with RA. Hepcidin is usually regulated by iron metabolism-iron overload (black arrow) leads to an increase in hepcidin, and iron deficiency (white arrow) leads to a decrease-and is suppressed by erythropoiesis, sex hormones, and growth factors. To maintain iron homeostasis, higher hepcidin levels result in a decrease in serum iron levels, while lower hepcidin levels increase serum iron levels. When inflammation due to RA occurs, production of hepcidin is increased through inflammatory cytokines that cause elevated ferritin levels. In this study, serum iron levels are positively related to serum hepcidin and ferritin levels but negatively related to inflammation due to RA. This unexpected relationship may give arisen because inflammation in most of the patients was well-controlled. Although iron overload and hepcidin may influence osteoporosis, in this study serum iron level was positively related to BMD but serum hepcidin level was not. However, serum 25(OH)D level was positively related to the serum hepcidin level and also positively related to the femoral Z score in this study. According to previous reports, opposite effects of serum iron level on BMD and 25(OH)D on hepcidin are indicated, and further research is needed to determine the mechanism. Serum FGF23 is not directly related to the serum hepcidin level, but serum 25(OH) D level and inflammation are common factors involved in the regulation of both serum hepcidin and FGF23 levels.