The impacts of hypoxia-inducible factor stabilizers on laboratory parameters and clinical outcomes in chronic kidney disease patients with renal anemia: a systematic review and meta-analysis

ABSTRACT

Renal anemia in chronic kidney disease (CKD) is associated with poor outcomes. Hypoxia-inducible factor (HIF) stabilizer, which induces endogenous erythropoietin synthesis and enhances iron mobilization, is a novel treatment for anemia in CKD. We conducted a systematic review and meta-analysis to analyze the effect of HIF stabilizers in anemic CKD patients. This meta-analysis included 43 officially published articles and 3 unpublished studies (27 338 patients). HIF stabilizer treatment significantly increased hemoglobin (Hb) level when compared with placebo (mean difference 1.19 g/dL; 95% confidence interval 0.94 to 1.44 g/dL; P < .001). There was no significant difference in Hb level when compared with erythropoiesis-stimulating agents (ESAs). Significant reductions of ferritin and transferrin saturation (TSAT) were observed, while total iron-binding capacity was increased in the HIF stabilizer group compared with placebo or ESAs. HIF stabilizers significantly reduced hepcidin, high-density lipoprotein, low-density lipoprotein and triglyceride levels. Acute kidney injury and thrombotic events were significantly observed in patients receiving HIF stabilizers. There were no significant differences in myocardial infarction, stroke, dialysis initiation, pulmonary hypertension and mortality between HIF stabilizer and control groups. The present meta-analysis provided evidence that HIF stabilizers increased Hb and TIBC levels and reduced hepcidin, ferritin and TSAT in CKD patients with renal anemia. Long-term follow-up studies on clinical outcomes of HIF stabilizers are still needed.

Graphical Abstract

Graphical Abstract.

INTRODUCTION

Anemia, one of the most common complications of chronic kidney disease (CKD), is mediated by insufficient erythropoietin (EPO) production from the kidney combined with absolute or functional iron deficiency [1]. In addition, chronic inflammation in CKD patients could contribute to EPO hyporesponsiveness and stimulate hepcidin, which affects iron homeostasis and exacerbates anemia in CKD [2]. The prevalence of renal anemia increases as CKD progresses and most end-stage kidney disease (ESKD) patients are affected by anemia [3]. The presence of anemia in CKD patients is associated with poor outcomes, including reduced quality of life, cognitive impairment, left ventricular hypertrophy, elevated hospitalization and increased mortality [4, 5]. Administration of erythropoiesis-stimulating agents (ESAs), such as recombinant human EPO (rHuEPO), coupled with iron supplementation is currently the mainstay of treatment for anemia in CKD. Although ESAs could effectively increase hemoglobin (Hb) levels and reduce the necessity for blood transfusion [6], recent studies have raised concerns that patients receiving exogenous ESAs to achieve normal Hb levels (≥13 g/dL) had higher cardiovascular events and greater mortality rates [7, 8]. Furthermore, Hb variability and excursion above the target range during treatment with exogenous ESAs might be associated with increased cardiovascular diseases [9]. Meanwhile, intravenous iron supplementation might increase the risk of an allergic reaction, inflammation and iron overload, [10, 11]. Therefore, a novel approach to anemia therapy that does not generate supraphysiologic EPO levels and reduces the need for iron supplementation should be developed.

Hypoxia-inducible factor (HIF) stabilizers are novel agents that inhibit HIF prolyl hydroxylase domain enzyme (HIF-PHD), leading to the activation of HIF signaling and transcription of thousands of HIF target genes, including EPO and iron metabolism genes [12]. These processes result in increased endogenous erythropoietin production and iron metabolism stimulation. Several different HIF stabilizers, such as roxadustat, daprodustat, vadadustat, molidustat, desidustat and enarodustat, have undergone investigation in various clinical trials. Earlier reports involved Phase 2 clinical trials while several Phase 3 clinical studies have been recently published and many more Phase 3 clinical trials are being released in the near future.

Meanwhile, there have been several meta-analyses of randomized controlled studies (RCTs) [13–17], most of which demonstrated that HIF stabilizers could provide beneficial effects on anemia-related parameters and could yield many positive pleiotropic effects in CKD patients. Nonetheless, certain meta-analyses assessed only one or few HIF stabilizers and one or few parameters [15, 16].

Despite the auspiciously reported outcomes of HIF stabilizers stated above, there are considerations regarding the safety issues following the long-term use of HIF stabilizers. Recent studies illustrated that HIF activation could stimulate various substances, such as transforming growth factor-beta (TGF-β) [18] and vascular endothelial growth factor (VEGF) [19], which could be potentially associated with the progression of renal function and cardiovascular diseases. All earlier meta-analyses provided insufficient data on HIF stabilizer–related clinical outcomes. In addition, most previous meta-analyses did not separately evaluate the effects of HIF stabilizers between non-dialysis-dependent (NDD)-CKD and dialysis-dependent (DD)-CKD patients [17]. Furthermore, assessment of the effects of individual HIF stabilizers has scarcely been explored.

The present meta-analysis was conducted to extensively evaluate the impacts of HIF stabilizers on laboratory parameters and clinical outcomes, including dialysis initiation, cardiovascular events and mortality in CKD patients with renal anemia. Officially published and unpublished studies up to March 2022 were included in the meta-analysis. The effects of overall and each HIF stabilizer were comprehensively examined in overall and NDD-CKD as well as DD-CKD patients.

MATERIALS AND METHODS

Data source and searches

The present work was conducted in accordance with the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement. The literature search was carried out in MEDLINE, Scopus and Cochrane Central Register of Controlled trials from 2010 to March 2022 to identify eligible studies. Furthermore, unpublished data were sought from ClinicalTrials.gov and the conference abstracts. The study protocol was registered with PROSPERO (CRD42022318794). Reference lists of the obtained articles were also searched for relevant publications. For the search, the following Medical Subject Heading terms were used: ‘HIF stabilizer’ OR ‘Hypoxia-Inducible Factor inhibitor’ OR ‘roxadustat’ OR ‘daprodustat’ OR ‘vadadustat’ OR ‘desidustat’ OR ‘molidustat’ OR ‘enarodustat’ OR ‘FG-4592’ OR ‘AKB-6548’ OR ‘GSK1278863’ OR ‘BAY85-3934’ OR ‘JTZ951’ OR ‘DS1093a’. The search was limited to English language and focused on all stages of CKD, including ESKD.

Study selection

We included studies if they were RCTs, sub-analyses of RCTs or observational cohorts (prospective or retrospective). Studies were required to meet the following inclusion criteria: (i) original article and (ii) intervention studies in humans with CKD which investigated the effects and/or safety of different anemia treatments in CKD patients. Interventions of interest included six kinds of HIF stabilizers (roxadustat, daprodustat, vadadustat, molidustat, desidustat and enarodustat) compared with control (ESAs as well as placebo). The outcomes included changes in Hb levels, iron parameters, lipid and inflammatory profiles from baseline and clinical outcomes, comprising myocardial infarction, stroke, acute kidney injury (AKI), dialysis initiation, abnormal liver function test, pulmonary hypertension (PHT), retinopathy, thrombotic events and mortality. Narrative reviews and case reports/series were excluded.

Two authors (K.T. and T.T.) independently screened the titles and abstracts of all electronic citations, and full-text articles were retrieved for the comprehensive review and were independently rescreened. Disagreements were resolved through consensus and arbitration by the third author (P.S.).

Data extraction and quality assessment

The following study characteristics and outcomes of interest were extracted and reviewed by independent authors: country of origin, year of publication, number of patients, mean age, population characteristics (NDD-CKD vs DD-CKD), treatment modality and duration of follow-up. For the studies with more than two intervention groups including multiple fixed dosages, each pair of comparisons was included in the analysis and considered as one study arm. The following outcomes of interest were evaluated: Hb, ferritin, transferrin saturation (TSAT), serum iron (SI), total binding capacity (TIBC), hepcidin, high-sensitivity C-reactive protein (hs-CRP), high-density lipoprotein (HDL) and low-density lipoprotein (LDL) levels at baseline and end-of-the study. Incidences of serious adverse events (SAEs), myocardial infarction, stroke, major adverse cardiovascular events (MACE), AKI, dialysis initiation, abnormal liver function test, PHT, retinopathy and mortality rates were also determined.

Risk of bias and GRADE assessment

Two reviewers (K.T. and T.T.) independently assessed the risk of bias for included studies using the Cochrane risk-of-bias tool for randomized trials (RoB2) [20]. The summary graphic was created using the RoB visualizing application RobVis22 to show the independent domains for risk of bias. The Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) [21] technique was used to assess the overall quality of the evidence. The overall quality of evidence in the included studies was categorized as either very low, low, moderate or high.

Data analysis

We used random-effects model meta-analysis to compute the mean difference (MD) from baseline to the end of the trial between intervention and control groups for continuous variables. The risk ratio (RR) was chosen to analyze binary variables, such as SAEs, myocardial infarction, stroke, AKI, dialysis initiation, abnormal liver function test, PHT, retinopathy and mortality rates. All pooled estimates are displayed with a 95% confidence interval (CI). The existence of heterogeneity was examined using the I² index and the Q-test P-value. An I² index >75% indicates medium to high heterogeneity.

To identify possible effect modifiers on the pooled analyses, sources of heterogeneity were explored by subgroup analyses according to population characteristics (NDD-CKD vs DD-CKD), control group (ESAs vs placebo) and types of HIF stabilizers. We conducted a sensitivity analysis to assess the consistency of the results. Meta-regressions were also explored to identify the causes of heterogeneity. Baseline of Hb and ferritin levels in HIF stabilizers and control group were investigated. Statistical significance was met when the P-value was <.05. Publication bias was formally assessed using funnel plots and the Egger test. The analyses were investigated using the Comprehensive Meta-Analysis software version 2.0 (www.meta-analysis.com; Biostat, Englewood, NJ, USA).

RESULTS

Search results and characteristics of the studies

A total of 2179 potentially relevant citations were identified from the aforementioned databases; 112 articles were retrieved for detailed evaluation, of which 46 fulfilled the eligibility criteria (43 officially published articles and 3 unpublished studies) [22–67]. A flow diagram of the study selection is exhibited in Fig. 1.

Figure 1:

PRISMA flow diagram.

The main characteristics of the studies are summarized in Table 1. The included studies were published between 2015 and March 2022. There were 29 855 patients with a mean age of 60.9 years. The HIF stabilizers studied in the present meta-analysis were roxadustat in 17 articles, daprodustat in 13 articles, vadadustat in 7 articles, enarodustat in 4 articles, molidustat in 4 articles and desidustat in 1 article. Twenty-one articles compared the efficacy of HIF stabilizers in NDD-CKD patients, while 20 articles did in DD-CKD patients. Five articles were performed in both NDD-CKD and DD-CKD patients (four articles divided patients into two studies by dialysis status (NDD-CKD and DD-CKD). Another article by Macdougall et al. which studied molidustat, divided patients into three studies: DIALOGUE 1, 2 and 4, which included patients with EPO-naïve NDD-CKD, NDD-CKD previously treated with ESAs, and DD-CKD patients who were previously treated with ESAs, respectively. Therefore, there were total 52 studies with 85 study arms in the final 46 eligible articles included in the meta-analysis. The follow-up duration ranged from 4 to 104 weeks.

Table 1:

Characteristics of the studies included in the meta-analysis.

Author	Nation	Study design	Sample size	Age (years)	Population	Intervention	Compare	Duration (weeks)
Besarab et al. (2015) [22]	USA	RCT	116	65.8	NDD-CKD	Roxadustat	Placebo	4
Provenzano et al. (2016) [23]	USA	RCT	144	56.7	Hemodialysis	Roxadustat	Epoetin alfa	19
Chen et al. (2017) [24]	China	RCT	91	60.2	NDD-CKD/dialysis	Roxadustat	Placebo, epoetin alfa	6
Akizawa et al. (2019) [25]	Japan	RCT	107	63.8	NDD-CKD	Roxadustat	Placebo	24
Chen et al. (2019) [26]	China	RCT	154	54.2	NDD-CKD	Roxadustat	Placebo	8
Chen et al. (2019) [27]	China	RCT	304	49	Dialysis	Roxadustat	Epoetin alfa	27
Akizawa et al. (2020) [28]	Japan	RCT	303	64.7	Hemodialysis	Roxadustat	Darbepoetin alfa	24
Akizawa et al. (2021) [29]	Japan	RCT	334	69.5	NDD-CKD	Roxadustat	Darbepoetin alfa	24
Barratt et al. (2021) [30]	Multiple countries	RCT	616	66.3	NDD-CKD	Roxadustat	Darbepoetin alfa	104
Charytan et al. (2021) [31]	USA	RCT	741	58	Hemodialysis	Roxadustat	Epoetin alfa	52
Coyne et al. (2021) [32]	USA	RCT	922	64.9	NDD-CKD	Roxadustat	Placebo	52
Csiky (2021) [33]	Multiple countries	RCT	836	61.4	DD-CKD	Roxadustat	ESAs	52
Fishbane et al. (2021) [34]	Multiple countries	RCT	2781	61.2	NDD-CKD	Roxadustat	Placebo	52
Hirai et al. (2021) [35]	Japan	Retrospective cohort	40	58.9	Peritoneal dialysis	Roxadustat	ESAs	24
Hou et al. (2021) [36]	China	RCT	129	48.1	Peritoneal dialysis	Roxadustat	ESAs	24
Provenzano et al. (2021) [37]	Multiple countries	RCT	1043	54	Dialysis	Roxadustat	Epoetin alfa	52
Shutov et al. (2021) [38]	Multiple countries	RCT	594	62.5	NDD-CKD	Roxadustat	Placebo	52
Brigandi et al. (2016) [39]	Multiple countries	RCT	107	56.14	CKD G3–5 and hemodialysis	Daprodustat	Placebo	4
Holdstock et al. (2016) [40]	Multiple countries	RCT	154	62.8	CKD G3–5 and hemodialysis	Daprodustat	Placebo	4
Akizawa et al. (2017) [41]	Japan	RCT	96	62.4	Hemodialysis	Daprodustat	Placebo	4
Bailey et al. (2019) [42]	Multiple countries	RCT	103	63.4	Hemodialysis	Daprodustat	Placebo	4
Holdstock et al. (2019) [43]	USA	RCT	252	66.1	NDD-CKD	Daprodustat	rHuEPO	16
Meadowcroft et al. (2019) [44]	Multiple countries	RCT	216	59.7	Hemodialysis	Daprodustat	rHuEPO	24
Akizawa et al. (2020) [45]	Japan	RCT	271	64	Hemodialysis	Daprodustat	Darbepoetin alfa	52
Nangaku et al. (2021) [46]	Japan	RCT	299	70	NDD-CKD	Daprodustat	Epoetin beta pegol	52
Singh et al. (2021) [47]	Multiple countries	RCT	3872	67	NDD-CKD	Daprodustat	Darbepoetin alfa	52
Singh et al. (2021) [48]	Multiple countries	RCT	2964	58.5	DD-CKD	Daprodustat	ESAs	52
NCT03409107 [49]	Multiple countries	RCT	614	N/A	NDD-CKD	Daprodustat	Placebo	28
NCT03029208 [50]	Multiple countries	RCT	312	N/A	Incident DD-CKD	Daprodustat	Darbepoetin alfa	52
NCT03400033 [51]	Multiple countries	RCT	407	N/A	DD-CKD	Daprodustat	Epoetin alfa	52
Pergola et al. (2016) [52]	USA	RCT	210	66.4	NDD-CKD	Vadadustat	Placebo	20
Martin et al. (2017) [53]	USA	RCT	93	65.5	CKD G3–4	Vadadustat	Placebo	6
Nangaku et al. (2020) [54]	Japan	RCT	109	66.8	NDD-CKD, dialysis	Vadadustat	Placebo	6
Chertow et al. (2021) [55]	Multiple countries	RCT	3476	66	NDD-CKD	Vadadustat	Darbepoetin alfa	52
Eckardt et al. (2021) [56]	Multiple countries	RCT	3923	57.9	Dialysis	Vadadustat	Darbepoetin alfa	52
Nangaku et al. (2021) [57]	Japan	RCT	304	72	NDD-CKD	Vadadustat	Darbepoetin alfa	52
Nangaku et al. (2021) [58]	Japan	RCT	323	65.5	DD-CKD	Vadadustat	Darbepoetin alfa	52
Akizawa et al. (2019) [59]	Japan	RCT	201	67.7	NDD-CKD	Enarodustat	Placebo	30
Akizawa et al. (2019) [60]	Japan	RCT	85	61.9	Hemodialysis	Enarodustat	Placebo	30
Akizawa et al. (2021) [61]	Japan	RCT	216	69.7	NDD-CKD	Enarodustat	Darbepoetin alfa	24
Akizawa et al. (2021) [62]	Japan	RCT	172	64	DD-CKD	Enarodustat	Darbepoetin alfa	24
Macdougall et al. (2019) [63]	Multiple countries	RCT	444	62.6	NDD-CKD, hemodialysis	Molidustat	rHuEPO	16
Akizawa et al. (2021) [64]	Japan	RCT	229	65.7	DD-CKD	Molidustat	Darbepoetin alfa	36
Yamamoto et al. (2021) [65]	Japan	RCT	164	70.7	ND-CKD	Molidustat	Darbepoetin alfa	52
Yamamoto et al. (2021) [67]	Japan	RCT	162	71.7	ND-CKD	Molidustat	Darbepoetin alfa	52
Parmar et al. [67]	India	RCT	117	48.1	NDD-CKD	Desidustat	Placebo	6

Risk-of-bias assessment

Supplement 1 in S1 File summarizes the results of the risk-of-bias assessment of the included studies. The overall risk of bias was judged to be low.

ΔHemoglobin

In meta-analysis of 85 study arms from 46 studies [22–67] which included 27 338 patients, HIF stabilizer treatment significantly increased Hb level compared with the control group (MD 0.659 g/dL; 95% CI 0.502 to 0.816 g/dL; P < .001) (Table 2). Focusing on the 16 studies with 37 study arms that examined the effects of HIF stabilizers and controllers on ESA-naïve patients and had Hb changes as an endpoint, HIF stabilizer therapy significantly increased Hb level compared with the control group (MD 0.95 g/dL; 95% CI 0.646 to 1.254 g/dL; P < .001).

Table 2:

Summary effects of HIF stabilizers compared with control group on anemia parameters, lipid profiles and inflammatory markers.

					Assessment of heterogeneity
Outcome variables	No. of study arms	No. of patients	Mean difference (95% CI)	P-value	I2 index	P-value	Assessment of publication bias Egger's test; P-value
Anemia parameters
Hb (g/dL)	85	27 338	0.659 (0.502, 0.816)	<.001	98.88	<.001	.805
Hepcidin (ng/mL)	68	20 270	−29.105 (−34.405, −23.805)	<.001	99.40	<.001	.289
Serum ferritin (ng/mL)	65	22 568	−42.797 (−57.003, −28.591)	<.001	99.43	<.001	<.001
Serum iron (μg/dL)	44	7844	2.153 (−0.441, 4.747)	.104	93.38	<.001	.114
TIBC (μg/dL)	58	9206	43.811 (39.158, 48.464)	<.001	96.28	<.001	.054
TSAT (%)	64	22 585	−2.999 (−3.672, −2.325)	<.001	98.04	<.001	.012
Lipid profiles
TC (mg/dL)	22	9847	−15.322 (−17.924, −12.720)	<.001	83.35	<.001	<.001
LDL (mg/dL)	21	13 370	−15.363 (−18.470, −12.257)	<.001	96.80	<.001	.54
HDL (mg/dL)	14	8634	−5.184 (−6.501, −3.866)	<.001	84.08	<.001	.007
TG (mg/dL)	10	1580	−20.307 (−28.456, −12.158)	<.001	0	.76	.78
Inflammatory marker
Hs-CRP (mg/L)	10	3332	0.010 (−0.201, 0.221)	.927	54.75	.019	.27

ΔHepcidin

In meta-analysis of 68 study arms from 34 studies [23–25, 29, 31–34, 36–41, 43–48, 52, 53, 56–67] which included 20 270 patients, there was a significant difference in Δhepcidin level after receiving HIF stabilizers compared with control treatment (MD −29.105; 95% CI −34.405 to −23.805; P < .001) (Table 2).

ΔSerum ferritin

In meta-analysis of 65 study arms from 38 studies [22–38, 40, 41, 43–48, 52, 53, 56–66] which included 22 568 patients, the MD of Δserum ferritin between patients treated with HIF stabilizers and the control group was −42.797 ng/mL (95% CI −57.003 to −28.591; P < .001) (Table 2).

ΔSerum iron

In meta-analysis of 44 study arms from 26 studies [22–24, 26–32, 34, 37, 40–46, 61–67] which included 7844 patients, there was no significant difference in Δserum iron between patients with HIF stabilizers and control group (MD 2.153 μg/dL; 95% CI −0.441 to 4.747; P = .104) (Table 2).

ΔTIBC

In meta-analysis of 58 study arms from 30 studies [22–29, 31, 32, 34, 37, 40–46, 52, 53, 59–67] which included 9206 patients, there was a significant increase of ΔTIBC between HIF stabilizers and control groups (MD 43.811 μg/dL; 95% CI 39.158 to 48.464 μg/dL; P < .001) (Table 2).

ΔTSAT

In meta-analysis of 64 study arms from 38 studies [22–38, 40–48, 56–67] which included 22 585 patients, the result showed a significant decrease ΔTSAT in patients with HIF stabilizers compared with control groups (MD −2.999; 95% CI −3.672 to −2.325; P < .001) (Table 2).

Lipid profiles

There were 22 study arms [23, 24, 27, 32, 36, 37, 40, 41, 47, 48], 21 study arms [24, 27, 30–32, 34, 37, 41, 46–48, 67], 14 study arms [24, 26, 36, 37, 41, 47, 48] and 10 study arms [24, 36, 37, 67] that reported the total cholesterol (TC), LDL, HDL and triglyceride (TG) outcomes, respectively. There were significant decreases in ΔTC (MD −15.322 mg/dL; 95% CI −17.924 to −12.720 mg/dL; P < .001), ΔLDL (MD −15.363 mg/dL; 95% CI −18.470 to −12.257 mg/dL; P < .001), ΔHDL (−5.184 mg/dL; 95% CI −6.501 to −3.866 mg/dL; P < .001) and ΔTG levels (MD −20.307 mg/dL; 95% CI −28.456 to −12.158 mg/dL; P < .001) between HIF stabilizers and control groups.

ΔSerum hs-CRP

In meta-analysis of 10 study arms from three studies [40, 48, 67] which included 3332 patients, there was no statistically significant difference in Δserum hs-CRP level between HIF stabilizers and control groups (MD 0.010 mg/L; 95% CI −0.201 to 0.221 mg/L; P = .927).

Incidence of SAEs

A total of 28 studies [23, 25–27, 29, 30, 32, 34, 36, 38, 42, 47, 48, 52, 53, 55–58, 61–67] comparing the incidence of SAEs between the HIF stabilizer and control groups were analyzed. Patients treated with HIF stabilizers had comparable risk of SAEs to the control group (RR 1.04; 95% CI 0.99 to 1.08; P = .053) (Table 3). However, HIF stabilizer–treated patients had slightly but a significantly higher incidence of AKI (RR 1.28; 95% CI 1.00 to 1.64; P = .049) (Fig. 2) and thrombotic events (RR 1.26; 95% CI 1.00 to 1.57; P = .047) (Fig. 3) than the control group. HIF stabilizers did not increase the risks of hypertension, hyperkalemia, abnormal liver function test, retinopathy and PHT (Table 3).

Table 3:

GRADE summary of findings: Effects of HIF stabilizers versus control group on clinical outcomes.

			Certainty assessment						Effect
Outcome variables	Study design	No. of studies	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	No. of patient	RR (95% CI)	Certainty
SAEs	RCT	28	Not serious	Not serious	Not serious	Not serious	None	22 602	1.04 (0.99, 1.08)	High ⊕⊕⊕⊕
Hypertension	RCT	30	Not serious	Not serious	Not serious	Not serious	None	25 156	0.99 (0.91, 1.09)	High ⊕⊕⊕⊕
Retinopathy	RCT	7	Serious	Not serious	Not serious	Serious	None	7 977	1.08 (0.82, 1.41)	Low ⊕⊕⊖⊖
Hyperkalemia	RCT	27	Not serious	Not serious	Not serious	Not serious	None	25 005	0.99 (0.88, 1.11)	High ⊕⊕⊕⊕
AKI	RCT	9	Serious	Not serious	Not serious	Serious	None	8 990	1.28 (1.00, 1.64)	Low ⊕⊕⊖⊖
Dialysis initiation	RCT	12	Not serious	Not serious	Not serious	Not serious	None	12 242	1.06 (0.95, 1.09)	High ⊕⊕⊕⊕
Abnormal liver function test	RCT	5	Not serious	Not serious	Not serious	Serious	None	1 234	1.30 (0.53, 3.22)	Moderate ⊕⊕⊕⊖
Thrombosis	RCT	11	Not serious	Not serious	Not serious	Serious	None	14 931	1.26 (1.00, 1.57)	Moderate ⊕⊕⊕⊖
PHT	RCT	6	Not serious	Not serious	Not serious	Serious	None	8 774	1.05 (0.61, 1.82)	Moderate ⊕⊕⊕⊖
Myocardial infarction	RCT	23	Not serious	Not serious	Not serious	Not serious	None	21 612	0.99 (0.87, 1.14)	High ⊕⊕⊕⊕
Stroke	RCT	19	Not serious	Not serious	Not serious	Not serious	None	19 496	0.97 (0.77, 1.23)	High ⊕⊕⊕⊕
MACE	RCT	12	Not serious	Not serious	Not serious	Not serious	None	16 472	1.00 (0.94, 1.07)	High ⊕⊕⊕⊕
Mortality	RCT	23	Not serious	Serious	Not serious	Not serious	None	22 113	0.91 (0.78, 1.07)	Moderate ⊕⊕⊕⊖

Figure 2:

Forest plot for incidence of AKI.

Figure 3:

Forest plot for thrombotic events.

Cardiovascular outcomes

There were no statistically significant differences in the incidences of myocardial infarction (RR 0.99; 95% CI 0.87 to 1.14; P = .927) (Fig. 4), stroke (RR 0.97; 95% CI 0.77 to 1.23; P = .818) (Fig. 5) and MACE (RR 1.00; 95% CI 0.94 to 1.07; P = .991) between HIF stabilizer and control groups (Table 3).

Figure 4:

Forest plot for incidence of myocardial infarction.

Figure 5:

Forest plot for incidence of stroke.

Dialysis initiation

Twelve studies [24, 26, 29, 30, 32, 34, 38, 47, 49, 52, 55, 62] which included 12 242 NDD-CKD patients reported the incidence of ESRD in patients receiving HIF stabilizers or control treatment. There was no significant difference in the incidence of dialysis initiation (RR 1.06; 95% CI 0.95 to 1.09; P = .672).

Mortality

Twenty-three studies [23, 29–31, 33, 36–38, 43, 44, 47–49, 52, 55–57, 59, 60, 63–66] including 22 113 patients reported mortality in patients treated with HIF stabilizers or control treatment. There was no statistically significant difference in mortality between HIF stabilizer and control groups (RR 0.91; 95% CI 0.78 to 1.07; P = .254) (Fig. 6).

Figure 6:

Forest plot for mortality.

Investigation of heterogeneity

In subgroup analysis according to population characteristics (NDD-CKD and DD-CKD), the differences in the outcomes of interest between HIF stabilizer and control groups were consistent with the overall population outcomes. ΔHb level was significantly higher in HIF stabilizer than control groups when studied in both NDD-CKD and DD-CKD patients (MD 0.817 g/dL; 95% CI 0.542 to 1.091; P < .001; and MD 0.268; 95% CI 0.191 to 0.345; P < .001, respectively). Also, there were significant decreased in Δhepcidin, Δserum ferritin and ΔTSAT between HIF stabilizer and control groups in both NDD-CKD and DD-CKD patients (Table 4). Nonetheless, HIF stabilizers provided significantly higher Δserum iron level only in DD-CKD patients (MD 8.971 μg/dL; 95% CI 5.474 to 12.468; P < .001). There was no significant difference in Δserum hs-CRP between HIF stabilizer and control groups in both NDD-CKD and DD-CKD patients.

Table 4:

Subgroup analysis of the effect of HIF stabilizers on anemia and lipid parameters by control group and population.

					Assessment of heterogeneity
Outcome variables	No. of study arms	No. of patients	MD (95% CI)	P-value	I2 index	P-value
Hb (g/dL)	85	27 338	0.659 (0.502, 0.816)	<.001	99.88	<.001
Population
NDD-CKD	47	14 233	0.817 (0.542, 1.091)	<.001	99.86	<.001
DD-CKD	38	13 105	0.268 (0.191, 0.345)	<.001	98.34	<.001
Control group
ESAs	37	20 362	−0.004 (−0.063, 0.055)	.899	98.4	<.001
Placebo	48	6976	1.190 (0.941, 1.440)	<.001	99.21	<.001
Type of HIF stabilizer
Roxadustat	19	8460	0.766 (0.403, 1.129)	<.001	99.65	<.001
Daprodustat	29	8236	0.665 (0.375, 0.955)	<.001	99.95	<.001
Vadadustat	14	8485	0.465 (0.251, 0.679)	<.001	94.99	<.001
Molidustat	6	882	0.205 (-0.192, 0.603)	.312	89.73	<.001
Desidustat	3	164	1.713 (0.945, 2.481)	<.001	56.84	.099
Enarodustat	14	1111	0.681 (0.139, 1.224)	.014	98.95	<.001
Hepcidin (ng/L)	68	20 270	−29.105 (−34.405, −23.805)	<.001	99.402	<.001
Population
NDD-CKD	41	9442	−32.777 (−39.974, −25.579)	<.001	99.481	<.001
DD-CKD	27	10 828	−24.290 (−30.181, −18.399)	<.001	96.932	<.001
Control group
ESAs	30	16 046	−19.117 (−25.500, −12.735)	<.001	99.002	<.001
Placebo	38	4224	−39.774 (−49.424, −30.124)	<.001	99.486	<.001
Type of HIF stabilizer
Roxadustat	16	5328	−15.254 (−26.880, −3.628)	.01	99.763	<.001
Daprodustat	22	8050	−48.230 (−56.715, −39.744)	<.001	92.368	<.001
Vadadustat	9	4904	−38.651 (−54.014, −23.288)	<.001	99.609	<.001
Molidustat	6	999	−9.630 (−21.035, 1.774)	.098	87.171	<.001
Desidustat	3	176	−34.769 (−56.649, −9.888)	.006	0	.533
Enarodustat	12	813	−14.808 (−20.485, −9.131)	<.001	92.822	<.001
Serum ferritin (ng/mL)	65	22 568	−42.797 (−57.003, −28.591)	<.001	99.43	<.001
Population
NDD-CKD	34	10 985	−43.204 (−60.470, −25.938)	<.001	99.616	<.001
DD-CKD	31	11 583	−40.516 (−72.914, −8.118)	.014	97.654	<.001
Control group
ESAs	33	17 072	−26.586 (−52.637, −0.535)	.045	98.132	<.001
Placebo	32	5496	−56.732 (−76.374, −37.089)	<.001	99.639	<.001
Type of HIF stabilizer
Roxadustat	22	7912	−53.453 (−98.872, −8.033)	.021	99.297	<.001
Daprodustat	16	7940	−32.525 (−46.895, −18.155)	<.001	69.155	<.001
Vadadustat	9	4904	−40.525 (−53.434, −27.617)	<.001	97.665	<.001
Molidustat	6	999	−8.554 (−42.822, 25.714)	.625	85.424	<.001
Desidustat	16	7940	−32.525 (−46.895, −18.155)	<.001	69.155	<.001
Enarodustat	12	813	−53.744 (−73.902, −33.585)	<.001	93.843	<.001
Serum iron (μg/dL)	44	7844	2.153 (−0.441, 4.747)	.104	93.382	<.001
Population
NDD-CKD	22	4463	−3.190 (−6.780, 0.399)	.082	96.035	<.001
DD-CKD	22	3381	8.971 (5.474, 12.468)	<.001	72.301	<.001
Control group
ESAs	24	4761	4.814 (0.962, 8.665)	.014	93.328	<.001
Placebo	20	3083	−1.444 (−5.610, 2.722)	.497	78.429	<.001
Type of HIF stabilizer
Roxadustat	17	5092	7.419 (4.089, 10.748)	<.001	95.187	<.001
Daprodustat	15	1193	−2.750 (−9.538, 4.037)	.427	84.551	<.001
Molidustat	6	999	−2.778 (−14.356, 8.800)	.638	92.12	<.001
Desidustat	3	176	−1.801 (−11.077, 7.474)	.703	0	.835
Enarodustat	3	384	−1.174 (−17.440, 15.091)	.887	95.792	<.001
TIBC (μg/dL)	58	9206	43.811 (39.158, 48.464)	<.001	96.287	<.001
Population
NDD-CKD	33	5711	41.891 (35.724, 48.058)	<.001	96.564	<.001
DD-CKD	25	3495	46.354 (37.086, 55.622)	<.001	95.222	<.001
Control group
ESAs	23	4063	27.781 (19.791, 35.771)	<.001	93.726	<.001
Placebo	35	5143	54.802 (48.352, 61.253)	<.001	96.806	<.001
Type of HIF stabilizer
Roxadustat	17	5704	44.681 (36.957, 52.406)	<.001	95.736	<.001
Daprodustat	15	1166	40.717 (29.011, 52.424)	<.001	89.262	<.001
Vadadustat	5	348	35.753 (21.198, 50.308)	<.001	97.423	<.001
Molidustat	6	999	6.526 (2.353, 10.699)	.002	0	.535
Desidustat	3	176	56.859 (34.341, 79.377)	<.001	28.457	.247
Enarodustat	12	813	64.730 (50.804, 78.656)	<.001	96.395	<.001
TSAT (%)	64	22 585	−2.999 (−3.672, −2.325)	<.001	98.045	<.001
Population
NDD-CKD	35	11 077	−4.687 (−5.819, −3.555)	<.001	98.81	<.001
DD-CKD	29	11 508	−0.799 (−1.736, 0.138)	.095	92.316	<.001
Control group
ESAs	33	17 048	−1.826 (−2.912, −0.739)	.001	98.102	<.001
Placebo	31	5537	−4.780 (−5.936, −3.623)	<.001	93.133	<.001
Type of HIF stabilizer
Roxadustat	22	8053	−1.115 (−1.709, −0.520)	<.001	91.917	<.001
Daprodustat	17	7994	−4.841 (−6.839, −2.844)	<.001	97.863	<.001
Vadadustat	4	4550	−1.718 (−4.859, 1.422)	.284	96.108	<.001
Molidustat	6	999	−1.520 (−5.937, 2.896)	.5	90.381	<.001
Desidustat	3	176	−5.434 (−9.201, −1.668)	.005	0	.544
Enarodustat	12	813	−5.856 (−8.825, −2.887)	<.001	90.774	<.001
TC (mg/dL)	22	9847	−15.322 (−17.924, −12.720)	<.001	88.355	<.001
Population
NDD-CKD	7	5005	−15.951 (−24.367, −7.536)	<.001	90.036	<.001
DD-CKD	15	4842	−16.667 (−20.114, −13.219)	<.001	83.32	<.001
Control group
ESAs	11	8326	−11.368 (−13.993, −8.743)	<.001	87.086	<.001
Placebo	1	1521	−20.242 (−27.365, −13.119)	<.001	83.225	<.001
Type of HIF stabilizer
Roxadustat	10	2677	−25.574 (−32.957, −18.191)	<.001	89.75	<.001
Daprodustat	12	7170	−9.760 (−12.457, −7.063)	<.001	70.85	<.001
LDL (mg/dL)	21	13 370	−15.363 (−18.470, −12.257)	<.001	96.803	<.001
Population
NDD-CKD	10	8120	−15.344 (−19.075, −11.613)	<.001	95.243	<.001
DD-CKD	11	5250	−15.080 (−20.744, −9.416)	<.001	90.898	<.001
Control group
ESAs	9	9572	−11.091 (−14.378, −7.803)	<.001	91.849	<.001
Placebo	12	3798	−19.215 (−23.004, −15.425)	<.001	66.068	.001
Type of HIF stabilizer
Roxadustat	11	6008	−16.811 (−19.172, −14.450)	<.001	66.023	.001
Daprodustat	7	7186	−6.364 (−9.153, −3.575)	<.001	78.306	<.001
Desidustat	3	176	−23.057 (−30.886, −15.229)	<.001	0	1
HDL (mg/dL)	14	8634	−5.184 (−6.501, −3.866)	<.001	84.082	<.001
Population
NDD-CKD	3	3993	−6.535 (−11.195, −1.874)	.006	81.028	.005
DD-CKD	11	4641	−5.779 (−7.936, −3.623)	<.001	84.755	<.001
Control group
ESAs	7	8125	−4.423 (−5.796, −3.050)	<.001	88.014	<.001
Placebo	7	509	−8.040 (−12.466, −3.614)	<.001	77.756	<.001
Type of HIF stabilizer
Roxadustat	8	1672	−6.090 (−8.768, −3.413)	<.001	80.817	<.001
Daprodustat	6	6962	−3.644 (−5.118, −2.171)	<.001	82.63	<.001
TG (mg/dL)	10	1580	−20.307 (−28.456, −12.158)	<.001	0	.759
Population
NDD-CKD	5	297	−28.582 (−41.781, −15.384)	<.001	0	.88
DD-CKD	5	1283	−15.209 (−25.568, −4.849)	.004	0	.7
Control group
ESAs	5	1283	−15.209 (−25.568, −4.849)	.004	0	.7
Placebo	5	297	−28.582 (−41.781, −15.384)	<.001	0	.88
Type of HIF stabilizer
Roxadustat	7	1404	−19.641 (−28.387, −10.896)	<.001	0	.58
Desidustat	3	176	−24.696 (−47.146, −2.245)	.031	0	.63

In subgroup analysis according to control groups (ESAs and placebo), there was no statistically significant change in Hb levels when compared with ESAs (MD −0.004 g/dL; 95% CI −0.063 to 0.055; P = .899). When compared with placebo, however, HIF stabilizer treatment significantly increased Hb level (MD 1.190 g/dL; 95% CI 0.941 to 1.440 g/dL; P < .001). The decrease in Δserum ferritin level between patients treated with HIF stabilizers compared with both placebo and ESAs was significant (MD –56.732; 95% CI –76.374 to −37.089; P < .001; and MD −26.586 ng/mL; 95% CI −52.637 to −0.535; P = .045). There was no significant difference in Δserum iron compared with placebo, whereas Δserum iron was significantly higher in patients treated with HIF stabilizers than ESAs (MD 4.814; 95% CI 0.962 to 8.665; P = .014). The significant decrease of TIBC in HIF stabilizers was observed when compared with placebo and ESA. Likewise, the effect of HIF stabilizers on hepcidin was similar between placebo and ESAs groups (MD −39.774; 95% CI −49.424 to −30.124; P < .001; and MD −19.117; 95% CI −25.500 to −12.735; P < .001, respectively) (Table 4).

With respect to the type of HIF stabilizers, patients treated with roxadustat, daprodustat, vadadustat, desidustat and enarodustat yielded significantly higher ΔHb levels and significantly lower Δserum ferritin and Δserum hepcidin, whereas there was no significant difference between molidustat and control groups. Only roxadustat group had a significant difference in Δserum iron compared with control groups.

To lessen the impact of Hb changes on iron parameters, the studies which aimed to maintain Hb levels in ESA-converted patients were analyzed. The iron study parameter changes between HIF stabilizer and control groups were consistent with the overall population outcomes. There were significant decreases in Δserum ferritin (MD −41.878 ng/mL; 95% CI −63.090 to −20.667 ng/mL; P < .001) and ΔTSAT levels (MD −3.374; 95% CI −4.290 to −2.457; P < .001), while a significant increase in ΔTIBC (48.006 μg/dL; 95% CI 37.508 to 58.505 μg/dL; P < .001) between HIF stabilizers and control groups. There was no significant difference in Δserum iron (MD 1.956 μg/dL; 95% CI −3.433 to 7.344 μg/dL; P = .477).

Regarding meta-regression, both baseline Hb and ferritin levels were not significantly associated with the risk of myocardial infarction, MACE and mortality rate.

Publication bias

As the Egger's test results were mainly insignificant (P > .05), together with a generally symmetrical funnel plot for the outcomes of the studies included in this meta-analysis, publication bias was less likely to occur.

DISCUSSION

The present meta-analysis extensively explored the efficacy and safety of HIF stabilizers in CKD patients with anemia. Forty-six articles, which included 52 studies, were identified. We assessed the changes in Hb levels, iron and inflammatory parameters, lipid profiles, adverse events and various clinical outcomes. We reported the significant differences in ΔHb levels, Δhepcidin, Δferritin, ΔTIBC, ΔTSAT, ΔTC, ΔLDL, ΔHDL and ΔTG between HIF stabilizer and control groups (Table 2).

Although there have been numerous meta-analyses regarding the effects of HIF stabilizers in CKD patients with renal anemia, only four meta-analyses, including three earlier studies and the present work, examined all six available HIF stabilizers simultaneously [13, 14, 17]. The present meta-analysis provided the most update-to-date and comprehensive data from all HIF stabilizers studies. Studies officially published in international medical journals and unpublished studies were included. The present study separately evaluated the effects of HIF stabilizers in NDD-CKD and DD-CKD patients; determined the effects of HIF stabilizers compared with controls (placebo or ESAs), placebo and ESAs; and assessed the effects of the individual type of HIF stabilizers. In addition, the present meta-analysis examined all components of lipid profiles and inflammatory markers. Moreover, vital clinical outcomes, comprising myocardial infarction, stroke, AKI, dialysis initiation, abnormal liver function test, PHT, retinopathy, thrombotic events and mortality, were attentively determined in the current study.

Renal anemia treatment with conventional ESA therapy could effectively reduce blood transfusion rate and improve quality of life in CKD patients; however, high doses of ESAs are associated with worsening hypertension and increased cardiovascular risk [7, 68]. Targeting the HIF pathway with HIF stabilizers has the potential to provide more physiologic endogenous EPO production [69]. Our meta-analysis showed that HIF stabilizers could promote erythropoiesis and increase Hb levels (Table 2). This effect persisted in both NDD-CKD and DD-CKD patients with anemia (Table 4). Among three previous meta-analyses, only the study by Wen et al. analyzed this issue and the results were supportive of our work. Regarding comparators, only the present study and the work by Wen et al. performed complete analyses, control (placebo or ESAs), placebo and ESA. The study by Wen et al., which enrolled 19 RCTs in 2020, revealed the positive effect of HIF stabilizers on Hb level compared with ESA or placebo [13]. After updating recent articles in 2020–21, our study showed that HIF stabilizers had a significantly superior effect only compared with placebo and not with ESAs. In the subgroup analysis by type of HIF stabilizers, five types of HIF stabilizers, except molidustat, could effectively raise the Hb level (Table 4). These results suggest that HIF stabilizers might have a class effect on erythropoiesis.

The elevation of hepcidin and ferroportin degradation in CKD patients leads to the reduction of intestinal iron absorption, impaired release of iron from internal stores and functional iron deficiency anemia. Besides promoting endogenous erythropoietin production, HIF upregulates transferrin receptor expression and promotes iron uptake via increased ferroportin transcription [70]. In addition, HIF activation could suppress hepcidin production and enhance iron uptake and transportation as well as increase the iron-binding capacity of transferrin [12]. Therefore, HIF stabilizers have been proposed to ameliorate all the iron metabolism abnormalities in CKD patients and avoid iron toxicity risk. The results from our meta-analysis established the lowering effects of HIF stabilizers on hepcidin and serum ferritin (Table 2). Also, the decreases in TSAT and the increase in TIBC indicated the effect of HIF stabilizers on iron utilization (Table 2).

Apart from erythropoiesis and iron metabolism, HIF also involves lipid metabolism. We revealed that HIF stabilizers effectively reduced TC, LDL and HDL levels (Table 2). These lowering effects could be explained via HIF-dependent effects on acetyl coenzyme A, an essential enzyme in the first step of cholesterol synthesis, and the degradation of 3-hydroxy-3-methylglutaryl coenzyme A reductase, the rate-limiting enzymes in cholesterol synthesis [71], respectively. A recent meta-analysis by Chen et al [17]. demonstrated only lowering effects of HIF stabilizers on TC and LDL, while the remaining two studies did not report this issue. The present study illustrated that HIF stabilizers reduced all lipid-related parameters, including TG and HDL (Table 2). Among these HIF stabilizers, the lipid-lowering effects of roxadustat and daprodustat were consistently reported. In contrast, the effect of vadadustat and molidustat on lipid metabolism has been observed infrequently.

Indeed, HIF pathway involves in the multiple gene regulation and production of VEGF, an angiogenic growth factor that could promote atherosclerosis, diabetic retinopathy, malignant tumor and progression of autosomal dominant polycystic progression kidney disease [72, 73]. Therefore, it should be considered that using HIF stabilizers might bring unfavorable outcomes. This meta-analysis showed no significant differences in mortality and cardiovascular events, including myocardial infarction, stroke and MACE (Table 3). However, most of the studies in this meta-analysis were performed in a non-inferiority design and had a short follow-up period. According to available evidence, HIF stabilizers are potentially beneficial therapy for renal anemia with a cardiovascular and mortality safety profile comparable but not superior to current ESA. Data regarding the beneficial effects of HIF stabilizers on cardiovascular and mortality outcomes are required in further studies. Although we found that HIF stabilizers yielded a significantly increased risk of AKI, there was no statistically significant risk of dialysis initiation. A recent study revealed the novel insight that HIF activation might lead to tubulointerstitial renal fibrosis through the connection between HIF and the TGF-β pathway [74]. Moreover, HIF is also related to VEGF and is potentially associated with the progression of renal diseases and diabetic retinopathy [72]. Therefore, despite the promising effects of HIF stabilizers on renal anemia, further studies focusing on adverse events, especially renal progression, are awaited.

This systematic review and meta-analysis enrolled the most updated published and unpublished data involving HIF stabilizers used for renal anemia therapy. We comprehensively analyzed the impact of HIF stabilizers on laboratory and clinical outcomes. In addition, we also performed several subgroup analyses among diverse populations, including NDD-CKD and DD-CKD. However, there were some limitations. Differences in patient characteristics, dose, frequency of HIF stabilizer application, and iron supplement of each study contributed to heterogeneity in the results and might affect the iron parameter outcomes. Additionally, several studies in NDD-CKD patients had a higher dropout rate in the placebo group compared with the HIF stabilizer group due to the requirement of ESA or rescue therapy in advanced NDD-CKD patients who progressed to ESKD requiring dialysis. This may affect the results of subsequent adverse events, especially cardiovascular events and mortality. As a result, despite the reassuring cardiovascular risk data from a recent study, the Food and Drug Administration voted against approving roxadustat in July 2021 because of cardiovascular safety concerns. Roxadustat met non-inferiority margins in an intention-to-treat analysis but not in an on-drug analysis, potentially due to significantly higher placebo arm dropout rates. Further large-scale RCTs examining the long-term follow-up on clinical outcomes of HIF stabilizers in CKD patients are needed.

CONCLUSION

The present meta-analysis provided evidence that HIF stabilizers increased Hb and TIBC levels in NDD-CKD and DD-CKD patients with renal anemia and reduced hepcidin, ferritin, serum iron and TSAT. HIF stabilizers also lowered TC, LDL, HDL and TG levels. Significant differences in cardiovascular events and mortality between HIF stabilizer and control groups were not identified in this study. Long-term follow-up studies on clinical outcomes of HIF stabilizers are still needed.

AUTHORS’ CONTRIBUTIONS

Conception and design: K.T., T.T. and P.S. Data collection: K.T., T.T., J.P. and P.S. Analysis and interpretation of the data: K.T., P.S., S.E. and K.P. Writing: K.T., T.T., P.S., S.E. and K.P. All authors read and approved the final manuscript.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.

CONFLICT OF INTEREST STATEMENT

All authors declare that there is no conflict of interest.