Predictive Value of Serum Hepcidin Levels for the Risk of Incident End-Stage Kidney Disease in Patients with Chronic Kidney Disease: The KNOW-CKD

Hepcidin is a peptide primarily produced by hepatocyte and in a small quantity by macrophages and adipocytes [1‒3]. The gene, hepcidin antimicrobial peptide (HAMP), is initially translated into an 84 amino acid pre-pro-hepcidin, which is post-translationally modified to a 60 amino acid pro-hepcidin, and finally to a mature C-terminal 25 amino acid peptide, hepcidin [4]. Beyond the discovery of hepcidin as an antimicrobial peptide in 2000 [1, 5], studies reported that mutations of HAMP lead to severe iron overload and hemochromatosis [6, 7]. Mechanistically, hepcidin binds to and induces degradation of ferroportin, a transmembrane iron exporter protein. As ferroportin is critical for iron absorption via duodenal cells and iron mobilization from the storage in macrophages and hepatocytes, hepcidin is the master regulator for systemic iron homeostasis, ultimately contributing to the development of anemia [8, 9].

Anemia is a common complication in the patients with chronic kidney disease (CKD), especially among those at advanced stages, affecting up to 60% of the patients with non-dialysis CKD [10, 11]. The pathogenesis of anemia in CKD is multifactorial, including decrease in erythropoietin (EPO) production due to reduced functional nephron mass, absolute and/or relative iron deficiency, and inflammation [8, 12]. Intriguingly, all of these are closely related to the function of hepcidin. As EPO suppresses the upregulation of hepcidin [13], reduced EPO production in patients with CKD results in elevated serum hepcidin levels. Inflammatory cytokines, such as interleukin-6 (IL-6), are well-known stimulators of hepcidin expression [14]. Further, the renal clearance of circulating hepcidin is also interfered with among the subjects with reduced glomerular filtration rates [15]. These all together promote elevation in serum hepcidin levels, leading to the development of anemia in CKD associated with inefficient iron metabolism [12]. Indeed, hepcidin is now acknowledged as a major culprit of the anemia in CKD, and novel therapeutics for renal anemia adopt serum hepcidin levels as a surrogate marker [16, 17].

However, despite the pivotal role in the development of anemia, the association between serum hepcidin levels and hard end points in CKD, such as initiation of renal replacement therapy, major adverse cardiac events, and all-cause mortality, is currently elusive. The prognostic value of serum hepcidin levels in combination of N-terminal pro-brain type natriuretic peptide (NT-proBNP) or neutrophil gelatinase-associated lipocalin (NGAL) has been previously evaluated in certain clinical contexts [18‒20] but not in patients with CKD. Therefore, here we investigated the association between serum hepcidin levels and the risk of end-stage kidney disease (ESKD) among the patients with pre-dialysis CKD.

The KNOW-CKD (KoreaN Cohort Study for Outcome in Patients With Chronic Kidney Disease) is a prospective cohort study that enrolled the patients with CKD at stages 1 to pre-dialysis 5 from nine tertiary hospitals in South Korea between 2011 and 2016 (NCT01630486 at http://www.clinicaltrials.gov) [21]. The study was designed and conducted according to the Declaration of Helsinki. The study protocol was reviewed and approved by the Institutional Review Board at each participating center, as previously described [22‒24]. All the participants voluntarily provided the informed consent. Each participating center closely monitored the participants during the follow-up period to report major events. The reported study outcomes were cross-checked by the collaborating investigators. Among a total of 2,238 patients who were initially recruited, after excluding those lacking the baseline measurement of serum hepcidin levels (n = 98), and those lacking the data on follow-up duration (n = 3), a total of 2,109 patients were finally included for the analyses (Fig. 1). The study observation period ended on March 31, 2022, for the median duration of 7.940 years.

The baseline data of the participants on demographics, anthropometrics, and medical history were recorded according to the study protocol [21]. Blood and urine samples were obtained following an overnight fasting and were analyzed in the central laboratory (LabGenomics, Seongnam, Korea). Estimated glomerular filtration rate (eGFR) was calculated by CKD Epidemiology Collaboration (CKD-EPI) equation using serum creatinine level, as previously described [25]. CKD stages were defined by eGFR according to the Kidney Disease: Improving Global Outcomes (KDIGO) guideline [26]. Echocardiographic examinations were conducted by the cardiologists at each participating center, who were blinded to the clinical data, according to standard approach [27].

Serum hepcidin levels were measured at the central laboratory using an ELISA (enzyme-linked immunosorbent assay) kit (DRG Diagnostics, Marburg, Germany) [21, 28], according to the manufacturer’s instructions. The intra- and inter-assay coefficients of variation were 2.1–9.9% and 11.5–14.6%, respectively. Due to the detection limit of 80 ng/mL at the maximum level, higher levels were recorded as 80 ng/mL [28].

The participants were categorized into quartiles by serum hepcidin levels: Q1, Q2, Q3, and Q4 (Fig. 1). The study outcome was incident ESKD, which included initiation of dialysis or kidney transplantation.

Continuous and categorical variables of the baseline characteristics by serum hepcidin levels were compared using one-way analysis of variance and the χ² test, respectively. The cumulative incidence of the study outcome was visualized by Kaplan-Meier survival curves and subsequently compared using the log-rank test. The last visiting date was used as the censoring date for those with follow-up loss. To determine the independent association between serum hepcidin levels and the risk of ESKD, Cox proportional hazard regression models were adopted. The models were adjusted for the potential confounders as the following: model 1 reported unadjusted hazard ratios (HRs); model 2 was adjusted for age and gender; model 3 was additionally adjusted for medical history, such as age-adjusted Charlson Comorbidity Index, primary cause of CKD, smoking status, and medications (e.g., angiotensin converting enzyme inhibitors and angiotensin receptor blockers, diuretics, statins, and antiplatelet/anticoagulant agents), and anthropometric data, including body mass index (BMI) and systolic blood pressure (SBP). Model 4 was finally further adjusted for the baseline laboratory data, including hemoglobin, albumin, high-density lipoprotein cholesterol (HDL-C), fasting glucose, 25-hydroxyvitamin D (25[OH]D), high-sensitivity C-reactive protein (hs-CRP), CKD stages, spot urine protein-to-creatinine ratio (PCR), left ventricular mass index, and left ventricular ejection fraction. Participants with any missing data were excluded from primary analyses. The results of Cox regression analyses were reported with HRs and 95% confidence intervals (CIs). The linear correlation between serum hepcidin levels (as a continuous variable) and the risk of ESKD was illustrated by the penalized spline curve. To prove the robustness of the primary findings, we planned sensitivity analyses. First, participants were re-categorized by serum hepcidin levels into tertiles and quintiles, instead of quartiles, for Cox regression analyses. Second, the cause-specific HRs for the study outcome was assessed, where the death events before reaching the study outcome were considered a competing risk and treated as censoring. Third, the model was further adjusted for the use of erythropoiesis-stimulating agents (ESAs) in addition to model 4. Lastly, any missing values in the primary analyses were replaced using multiple imputation to repeat the Cox regression analyses. Pre-defined subgroup analyses were also implemented to examine if the association between serum hepcidin levels and the risk of ESKD is differed across specific clinical settings. These subgroups were defined by age (either <60 or ≥60 years), gender (male or female), BMI (<23 or ≥23 kg/m²), eGFR (either <45 or ≥45 mL/min/1.73 m²), and spot urine ACR (either <300 or ≥300 mg/g). Two-tailed p values below 0.05 were deemed statistically significant. All statistical analyses were executed using SPSS for Windows version 22.0 (IBM Corp., Armonk, NY, USA) and R software (version 4.1.1; R Project for Statistical Computing, Vienna, Austria).

The baseline characteristics of the study participants were described by serum hepcidin levels (Table 1). Higher serum hepcidin levels were significantly related to shorter duration of follow-up, older ages, higher frequency of male gender, higher Charlson Comorbidity Index score, higher frequency of diabetes mellitus as a primary cause of CKD, higher frequency of smoking history, more prevalent use of diuretics, statins, iron replacement, and ESAs, and higher SBP. Higher serum hepcidin levels were also related to lower hemoglobin, transferrin saturation, albumin, total cholesterol, HDL-C, and 25(OH)D levels and to higher ferritin, triglycerides, spot urine PCR, and creatinine levels. Accordingly, higher serum hepcidin levels were significantly associated with more advanced CKD stages of the participants. In the echocardiographic examination, left ventricular mass index, E/e’ (the ratio of the early transmitral blood flow velocity to early diastolic velocity of the mitral annulus), left atrium diameter, posterior wall thickness, interventricular wall thickness, and left ventricular end-diastolic diameter positively correlated with serum hepcidin levels (online suppl. Table S1; for all online suppl. material, see https://doi.org/10.1159/000542057). In the overall, higher serum hepcidin levels were in general associated with more severe comorbid conditions at the baseline.

Kaplan-Meier survival curve analysis visualized that the cumulative incidence of ESKD was significantly differed by serum hepcidin levels (p < 0.001, by log-rank test), with the highest incidence in Q4 (Fig. 2). Cox proportional hazard models to address the independent association between serum hepcidin levels and the risk of incident ESKD demonstrated that, compared to Q1, the risk of incident ESKD is significantly increased in Q4 (adjusted HR 1.372, 95% CI: 1.070–1.759) (Table 2). Every 1 log increase in serum hepcidin levels was associated with significant elevation of the risk of incident ESKD (adjusted HR 1.125, 95% CI: 1.031–1.228). Penalized spline curve analysis illustrated a linear, positive correlation between serum hepcidin levels and the risk of incident ESKD (Fig. 3).

Upon re-categorizing participants by serum hepcidin levels into tertiles or quintiles, compared to the 1st tertile and quintile, the risk of incident ESKD was significantly higher in the 3rd tertile (adjusted HR 1.265, 95% CI: 1.013–1.579) and in the 5th quintile (adjusted HR 1.481, 95% CI: 1.124–1.951) (online suppl. Table S2). After censoring the death events prior to the onset of ESKD, the risk of incident ESKD was still significantly increased in Q4, compared to Q1 (adjusted HR 1.373, 95% CI: 1.041–1.810) (Table 3). Additional adjustment for the use of ESAs did not significantly alter the primary result (adjusted HR for Q4 1.344, 95% CI: 1.045–1.729) (online suppl. Table S3). Even, after replacing missing values by multiple imputation methods, the risk of incident ESKD remained significantly increased in Q4, compared to Q1 (adjusted HR 1.289, 95% CI: 1.019–1.631) (Table 4). Subgroup analyses demonstrated that the association between serum hepcidin levels and the risk of incident ESKD was significant in the subjects with eGFR <45 mL/min/1.73 m² but not in those with eGFR ≥45 mL/min/1.73 m² (p for interaction = 0.023). The other variables, such as age, gender, BMI, and albuminuria, did not significantly modified the association between serum hepcidin levels and the risk of incident ESKD (Table 5).

In the present study, we discovered that elevation in serum hepcidin levels is significantly associated with increased risk of incident ESKD among the patients with pre-dialysis CKD. We believe that the robustness of the finding is convincing because the sensitivity analyses, including competing risk analysis and multiple imputation, yielded the comparable result to that of the primary analysis. Furthermore, subgroup analyses suggest that the association is significantly more prominent in the patients with advanced CKD (eGFR <45 mL/min/1.73 m²).

The vast majority of the studies on hepcidin have highlighted its biological function as an inhibitor of ferroportin [29, 30] or as an antimicrobial peptide [31, 32]. Yet the clinical implication of serum hepcidin levels on the overall prognosis has been largely neglected. Only a few studies focused on the prognostic value of serum hepcidin levels [18‒20], none of which evaluated the potential as a predictor of CKD progression. In addition, the previous studies analyzed the predictive role of hepcidin in combination of other biomarkers, such as NGAL and NT-proBNP [18‒20]. Therefore, it is a fining of novelty that elevation in serum hepcidin levels is independently associated with risk of incident ESKD among the patients with pre-dialysis CKD.

It is still uncertain whether hepcidin directly result in kidney damage in the natural course of CKD. Deletion of Hamp gene obviously alleviated anemia and prevented growth retardation in a murine model of adenine-induced CKD [33], while the comparison of kidney functions did not demonstrate any significant differences between wild type and knockout mice. Similarly, even though the overexpression of hepcidin in a human kidney cell line increased the intracellular iron contents, no definitive data of cellular damage such as cell death and fibrosis have been presented [34]. Thus, currently available evidence does not support that hepcidin directly accelerates CKD progression. On the other hand, it does not seem likely either that hepcidin may play a compensatory role against CKD progression. Although hepcidin demonstrated a kidney-protective role in a murine model of experimental urinary tract infection [32], the context should be apart from CKD. Collectively, it is postulated that the functional role of hepcidin in the progression of CKD is minor and is mostly limited to the development of renal anemia.

Nevertheless, based on our finding, the elevation in serum hepcidin levels is an independent predictor of CKD progression. Although hepcidin critically contributes to the development of renal anemia by alteration of iron metabolism, the expression of hepcidin is regulated by diverse factors, one of which is inflammation [8, 9, 12]. IL-6, by activation of STAT3 (signal transducer and activator of transcription 3), directly upregulates the transcription of hepcidin [14, 35, 36]. Alternatively, it is also well known that the reduced renal production of EPO in patients with CKD results in unopposed hepcidin expression [37]. Because both inflammation and impaired EPO production are remarkable features of advanced CKD, it seems reasonable that the elevation in serum hepcidin levels is a reflection of deteriorating kidney function and may predict the risk of incident ESKD.

Several limitations are to be acknowledged in the current study. First, due to the observational study design, we are not able to confirm the casual relation between serum hepcidin levels and the risk of incident ESKD. Indeed, as hepcidin is a major pathogenic molecule of renal anemia, and as the severity and prevalence of anemia are increased among the patients with more advanced CKD, serum hepcidin levels may be simply elevated in the patients with more severe renal anemia and low eGFR. Yet we rigorously adjusted hemoglobin levels in Cox proportional hazard models to prove the association between serum hepcidin levels and a kidney outcome that is independent of the presence or absence of anemia. Therefore, it is very likely that serum hepcidin levels are a predictor for the risk of incident ESKD. Second, the mechanistic insights on the relation between serum hepcidin levels and CKD progression are lacking in the current study, which mainly analyzed clinical data. However, we included a thorough review on the currently available publications on hepcidin, which strongly indicates a potential role of serum hepcidin levels as a predictor of CKD progression, as discussed above. Third, because the KNOW-CKD is a cohort of Korean patients resident in South Korea, the extrapolation of the result presented in the current study requires a precaution. Nevertheless, provided that the association between serum hepcidin levels and renal prognosis in patients with pre-dialysis CKD has barely been studied so far, the cross-validation of the finding using other CKD cohorts should be required. Fourth, serum hepcidin level was measured only once at the baseline in the current study, while serum hepcidin levels may increase as CKD progresses, as the eGFR significantly decreases as serum hepcidin levels increase (Table 1). Yet, based on this correlation, we assume that the repeated measurement of serum hepcidin levels may more precisely predict the risk of incident ESKD, even though only a single measurement was significantly associated with renal prognosis with the adjustment of the baseline status of kidney function.

In conclusion, we report that elevation in serum hepcidin levels is significantly associated with increased risk of incident ESKD among the patients with pre-dialysis CKD. It is expected that the measurement of serum hepcidin levels may help early identification of the patients at higher risk of CKD progression.

KNOW-CKD: Study Group Clinical Centers: Seoul National University: Curie Ahn, Kook-Hwan Oh, Dong Wan Chae, Ho Jun Chin, Hayne Cho Park, Seungmi Lee, Hyun Hwa Jang, and Hyun Jin Cho; Yonsei University and Severance Hospital: Kyu Hun Choi, Seung Hyeok Han, Tae Hyun Yoo, and Mi Hyun Yu; Kangbuk Samsung Medical Center: Kyubeck Lee and Sooyeon Jin; the Catholic University of Korea and Seoul St. Mary’s Hospital: Yong-Soo Kim and Sol Ji Kim; Gachon University and Gil Hospital: Wookyung Chung, Youkyoung Jang, and Ji Hye Park; Eulji University and Eulji General Hospital: Young-Hwan Hwang, Su-Ah Sung, and Jeong Ok So; Chonnam University: Soo Wan Kim, MD, and Ji Seon Lee; Inje University and Pusan Paik Hospital: Yeong Hoon Kim, Sun Woo Kang, and Yun Jin Kim; Epidemiology and Biostatistics Department of Preventive Medicine and Seoul National University College of Medicine: Byung-Joo Park, Sue Kyung Park, and Juyeon Lee; Coordinating Center, Medical Research Collaborating Center, Seoul National University Hospital, and Seoul National University College of Medicine: Joongyub Lee, Dayeon Nam, Soohee Kang, and Heejung Ahn; Central Laboratory: Donghee Seo, LabGenomics, Korea, and Dae Yeon Cho, LabGenomics, Korea; Biobank: Korea Biobank and Korea Centers for Disease Control and Prevention: Osong, Korea; and Korea Centers for Disease Control and Prevention: Dukhyoung Lee, Hy-ekyung Park (Project Officer), Eunkyeong Jung (Project Officer), Suyeon Jeong, Eunmi Ahn, and Sil-Hea Sung.

The study protocol of the KNOW-CKD was reviewed and approved by the Institutional Review Boards of the participating centers (Seoul National University Hospital [1104–089-359, May 25, 2011], Seoul National University Bundang Hospital [B-1106/129–008, August 24, 2011], Yonsei University Severance Hospital [4–2011-0163, June 2, 2011], Kangbuk Samsung Medical Center [2011–01-076, June 16, 2012], Seoul St. Mary’s Hospital [KC11OIMI0441, June 30, 2011], Gil Hospital [GIRBA2553, August 8, 2011], Eulji General Hospital [201105–01, June 10, 2011], Chonnam National University Hospital [CNUH-2011-092, July 5, 2011], and Busan Paik Hospital [11–091, July 26, 2011]). Patients only who voluntarily gave informed consent were enrolled.

The authors have no conflicts of interest to declare.

This work was supported by the Research Program funded by the Korea Disease Control and Prevention Agency (2011E3300300, 2012E3301100, 2013E3301600, 2013E3301601, 2013E3301602, 2016E3300200, 2016E3300201, 2016E3300202, 2019E320100, 2019E320101, 2019E320102, and 2022-11-007), by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (RS-2023-00217317 and RS-2023-00278258), and by the Chonnam National University Hospital Biomedical Research Institute (BCRI24046).

Conceptualization, statistical analysis, and original draft: S.H.S. Methodology: H.S., T.R.O., H.S.C., C.S.K., E.H.B, and S.K.M. Data collection and analysis: K.-H.O., K.-B.L., and J.Y.J. Supervision and funding acquisition: K.-H.O. and S.W.K. All authors provided critical feedback and helped shape the research, analysis, and manuscript.

The data presented in this study are available on request from the corresponding author.