Efficacy and safety analysis of different treatment schemes of erythropoiesis-stimulating agents for chronic kidney disease with anemia: a meta analysis between 2000–2024

Abstract

Objective

This study aims to evaluate the efficacy and safety of different regimens of erythropoiesis-stimulating agents (ESAs) in the treatment of anemia associated with chronic kidney disease. The regimens analyzed include a low-frequency high-dose epoetin group, a high-frequency low-dose epoetin group, and a Darbepoetin group.

Methods

A comprehensive search was conducted across several databases, including the China National Knowledge Infrastructure (CNKI), Wanfang Database, China Biology Medicine disc (CBM), PubMed, Web of Science, and Embase. We focused on randomized controlled trials (RCTs) related to ESA treatment for renal anemia published between January 1, 2000, and August 31, 2024. Data were extracted from studies meeting the inclusion criteria, and the Cochrane risk of bias assessment tool was employed for analysis. A meta-analysis of hemoglobin levels and adverse events was performed using RevMan 5.4.

Results

Twenty-eight studies were included in this review. Of these, 22 studies analyzed the frequency and efficacy of various dosage forms of short-acting ESAs, encompassing 1,670 patients with renal anemia. Additionally, 6 studies comprising 282 patients evaluated the efficacy and safety of Darbepoetin compared to high-frequency low-dose short-acting ESAs. The efficacy analysis revealed no statistically significant differences in hemoglobin level increases between the low-frequency high-dose group and the high-frequency low-dose group of epoetin [MD = 1.38, 95% CI (−0.64, 3.40), Z = 1.34, p > 0.05], nor between the Darbepoetin and short-acting ESA groups [MD = 0.23, 95% CI (−0.16, 0.62), Z = 1.16, p > 0.05]. In terms of safety, there were also no statistically significant differences between the low-frequency high-dose group and the high-frequency low-dose group [MD = 1.30, 95% CI (0.98, 1.73), Z = 1.82, p > 0.05]; however, the Darbepoetin group demonstrated a superior safety profile compared to the short-acting ESA group [MD = 0.63, 95% CI (0.41, 0.97), Z = 2.09, p < 0.05].

Conclusion

The results indicate no significant difference in efficacy and safety between the low-frequency and high-frequency groups. Nevertheless, the low-frequency administration of erythropoietin (Darbepoetin) resulted in fewer adverse reactions, potentially improving patient compliance.

Clinical trial number

Not applicable.

Renal anemia is a common complication in patients with chronic kidney disease (CKD) that significantly affects their quality of life and treatment outcomes; it is also positively correlated with the incidence and mortality of cardiovascular disease. Renal anemia primarily stems from insufficient secretion of erythropoietin (EPO) by the kidneys, coupled with iron metabolism disorders, accumulation of uremic toxins, oxidative stress, and inflammation. The administration of exogenous erythropoiesis-stimulating agents (ESAs) has led to improvements in the management of renal anemia. Recombinant human erythropoietins (rHuEPO) stimulate the growth and differentiation of bone marrow erythroid precursor cells by binding to the EPO receptor, thereby alleviating anemia symptoms and reducing transfusion requirements in CKD patients. Various EPO formulations and dosing regimens exist in clinical practice, with typical approaches including smaller doses administered two to three times per week or larger doses given once weekly. The long-interval (weekly) dosing regimen can alleviate injection site pain and enhance patient compliance. This study systematically evaluates the efficacy and safety of existing research on different formulations and dosing intervals, aiming to provide evidence-based recommendations for clinical practice.

Introduction

Data sources and search strategy

The study has been registered with the Prospective Register of Systematic Reviews (PROSPERO) under ID: CRD42024596526. The search terms utilized were: (erythropoietin OR recombinant human erythropoietin OR erythropoiesis-stimulating agent OR epoetin OR darbepoetin) AND (renal anemia OR anemia of chronic kidney disease OR dialysis anemia) AND (randomized OR randomized controlled trial OR RCT). The databases searched included the China National Knowledge Infrastructure (CNKI), Wanfang Database, and China Biology Medicine Disc (CBM). Additionally, in PubMed, Web of Science, and Embase, the search terms (EPO OR r-HuEPO OR epoetin OR darbepoetin) AND (renal anemia OR CKD) were employed. The search timeframe extended from January 1, 2000, to August 31, 2024.

Inclusion Criteria

1. Patients diagnosed with renal anemia (age ≥ 18 years, glomerular filtration rate < 90 mL/min/1.73 m², Hb < 12 g/dL).

2. All included studies are randomized controlled trials (RCTs), regardless of whether blinding or allocation concealment was employed.

3. The experimental group received high-dose ESAs treatment once a week (QW); the control group received low-dose ESAs treatment two or three times a week (BIW/TIW).

4. Hemoglobin (Hb) was used as the clinical efficacy evaluation indicator.

Exclusion Criteria

1. Non-randomized controlled trials, non-clinical trial studies.

2. Studies with inconsistent research content or inconsistent primary outcome measures.

3. Duplicate publications or studies with overlapping data, retaining only one publication.

4. Studies for which the full text could not be obtained.

Assessment of literature quality

Two reviewers independently assessed the risk of bias for each included study using the Cochrane Risk of Bias 2.0 (RoB 2.0) tool for randomized trials [1]. Each domain was categorized as “low risk,” “some concerns,” or “high risk.” Discrepancies were resolved through consensus or consultation with a third reviewer. We did not convert the risk-of-bias judgments into an overall “quality score.” The specific evaluation items included:

1. Randomization sequence generation;

2. Allocation concealment;

3. Blinding of participants and personnel (performance bias);

4. Blinding of outcome assessment (detection bias);

5. Completeness of outcome data;

6. Selective reporting;

7. Other biases.

Two reviewers independently assessed the risk of bias for each included trial using the Cochrane Risk of Bias 2.0 (RoB 2.0) tool, evaluating the five domains of “randomization process,” “deviations from intended interventions,” “missing outcome data,” “measurement of the outcome,” and “selection of the reported result.” Disagreements were resolved through consultation with a third methodologist. Studies with unclear random-sequence generation or allocation concealment were rated as “some concerns,” while those utilizing coin toss or odd-even allocation were judged as “high risk.” Open-label trials with objective hemoglobin outcomes received a rating of “some concerns” for the intervention domain and “low risk” for the measurement domain. A sensitivity analysis was prespecified to exclude trials with three or more domains rated as “high risk.” Among the 22 short-acting erythropoiesis-stimulating agent (ESA) trials, none reached three or more “high-risk” domains. After excluding two high-risk studies, the pooled mean difference (MD) decreased slightly from 1.38 to 0.97 g/L (95% confidence interval still crossing zero), confirming the robustness of the main conclusion. For the six darbepoetin trials, the principal limitation was inadequate randomization details; the exclusion of one trial with more than three “some concerns” domains did not materially affect the estimate. Funnel-plot Egger’s test yielded a p-value of 0.04. Trim-and-fill imputation added two virtual studies, and the adjusted effect remained clinically nonsignificant, indicating that publication bias had limited influence on the findings.

Statistical methods

Meta-analysis was conducted using RevMan 5.4 software [2]. For continuous variables, the effect size was represented by the mean difference (MD) or standardized mean difference (SMD), while for binary variables, the odds ratio (OR) was utilized. Each effect size was reported with a point estimate and a 95% confidence interval (CI). Heterogeneity among studies was assessed using the p-value and the I² statistic: a p-value greater than 0.1 and an I² value less than 50% indicated low or no heterogeneity, thereby warranting the application of a fixed-effects model. Conversely, a p-value of 0.1 or less, or an I² value greater than 50%, indicated significant heterogeneity, prompting a sensitivity analysis to identify its sources. If sources of heterogeneity were identified and addressed, and if heterogeneity remained low or diminished, the fixed-effects model was employed; otherwise, a random-effects model was applied. Funnel plots were utilized to analyze publication bias.

In addition to baseline hemoglobin (Hb) concentration, the included trials were balanced within each comparison regarding the route of administration (intravenous vs. subcutaneous), iron supplementation (oral vs. parenteral vs. none), and other potential confounders. Meta-regression revealed no significant influence of these variables on the Hb response (all p > 0.10); therefore, no further adjustments were made.

Results

Literature screening results

The study selection process is summarised in Fig. 1.

Fig. 1.

PRISMA 2020 flow diagram of study selection process

Basic characteristics of included literature

A total of 28 studies were included [3–30], in the analysis, of which 22 studies [3–24] focused on the frequency of administration and the efficacy of various doses of short-acting erythropoiesis-stimulating agents (ESAs), involving 1670 patients with renal anemia. Additionally, 6 studies [25–30]examined the efficacy of darbepoetin compared to high-frequency, low-dose short-acting ESAs, including a total of 282 patients with renal anemia (Table 1). Among the 22 included trials, the majority (n = 17) were Chinese studies that either stated “no obvious adverse events” or did not mention safety outcomes. Only one Chinese trial [14] reported three cases of hypertension (two in the intervention arm and one in the control arm). The four English-language randomized controlled trials (RCTs) declared similar safety profiles between dosing regimens but provided no extractable numerical data. Consequently, a formal meta-analysis of adverse events could not be performed, and findings are presented descriptively.We therefore adopted “total adverse-event (AE) episodes per treated subject” as a surrogate safety metric.

Table 1.

Basic characteristics of included literature

Researcher	Sample size		Intervention measures
	Experimental group	Control group	Experimental group	Control group
XiuLing Wu [14]	29	30	6000IU, qw，SC	2000IU, tiw，SC
ZhenHua Ge[13]	55	55	10000IU, qw，SC	3500 IU, tiw，SC
LuoJia Jiang [12]	60	60	10000IU, qw，SC	5000IU, tiw，SC
XiuWen Li [11]	35	34	10000IU, qw，SC	3000IU, biw or tiw，SC
Yang Qiu [10]	30	30	10000IU, qw，SC	3000IU, tiw，SC
TongQing Huang [9]	30	30	10000IU, qw，SC	3000IU, biw+4000, qw，SC
YunXi Xiang [8]	30	30	10000IU, qw，SC	3000IU, tiw，SC
Min Xiu [7]	27	27	10000IU, qw，SC	3000IU, tiw，SC
HuiWen Hao [5]	20	23	10000IU, qw，SC	3000IU, biw or tiw，SC
ShaoTing Wang [6]	33	29	10000IU, qw，SC	3000IU, biw or tiw，SC
Yan Chen [4]	15	15	10000IU, qw，SC	3000IU, tiw，SC
ZanWei Huang [3]	23	23	10000IU, qw，SC	3000IU, biw or tiw，SC
QiSheng Zhou [15]	38	35	6000IU, qw，SC	3000IU, biw，SC
LaMei Xiu [16]	18	18	10000IU, qw，SC	3000IU, biw，SC
DeYue Feng [17]	16	13	9000IU, qw，SC	3000IU, tiw，SC
Ping Zhang [18]	30	30	9000IU, qw，SC	3000IU, tiw，SC
1SuQin Gao [19]	10	10	10000IU, qw，SC	2000IU, tiw，SC
Jun Wu [20]	10	10	10000IU, qw，SC	3000IU, biw or tiw，SC
Pablo E. Pergola [21]	124	121	10000IU, qw	50 IU/kg, tiw
Young Ki Lee [22]	40	35	High dose, qw，SC	Low dose, biw or tiw，SC
Gabriel Mircescu [23]	102	101	High dose, qow，SC	Low dose, qw，SC
Lars G. Weiss [24]	88	30	High dose, qw，SC	Low dose, biw or tiw，SC
			darbepoetin alfa	short-acting ESAs
Jian Xiong [25]	17	19	40 μg, qw，IV	4000 IU/, tiw，IV
Nan Chen [26]	21	14	20 μg, qw，IV	3000IU, biw or tiw，IV
Mehta, K S [27]	21	20	0.45 μg/kg, qw，IV	50 IU/kg, tiw，IV
Shubhadeep D. Sinha [28]	55	51	0.45 μg/kg, qw，IV	50 IU/kg, tiw，IV
Wen-Yi Li [29]	18	19	2–4 times per month (1 μg = 200 IU)，IV	5–10 times per month，IV
Michael Allon [30]	13	14	qow (1 μg = 200 IU)，IV	qw，IV

The frequency of administration and the efficacy of various doses of short-acting erythropoiesis-stimulating agents (ESAs) can influence hemoglobin concentrations. However, other factors can also affect this response, such as the route of administration (intravenous vs. subcutaneous) or the type of iron supplementation (oral, parenteral, or none). Therefore, it is essential to evaluate the potential influence of multiple factors on hemoglobin concentrations(Table 2).

Table 2.

The literature of factors that may influence hemoglobin concentrations

Researcher	iron regimen	baseline Hb/SF/TSAT
		Experimental group		Control group
ZanWei Huang [3]	iron supplementation	Hb 63.57 ± 8.92 g/L; SF & TSAT not reported		Hb 65.07 ± 9.98 g/L; SF & TSAT not reported
Yan Chen [4]	Oral iron	Hb 64.8 ± 8.4 g/L; SF & TSAT not reported		Hb 65.25 ± 8.43 g/L; SF & TSAT not reported
HuiWen Hao [5]	Intravenous iron	Hb 49.1 ± 5.7 g/L g/L; SF & TSAT not reported		Hb 48.8 ± 4.2 g/L g/L; SF & TSAT not reported
ShaoTing Wang [6]	Intravenous iron	Hb 62.28 ± 8.35 g/L;TSAT 20%;SF not reported		Hb 61.48 ± 9.42 g/L;TSAT 20%; SF not reported
Min Xiu [7]	Intravenous iron(10 mg iron sucrose, once weekly)	Hb 70.05 ± 6.24 g/L;SF & TSAT not reported		Hb 70.02 ± 7.02 g/L;SF & TSAT not reported
YunXi Xiang [8]	Oral iron	Hb 84.4 ± 8.4 g/L;SF 13.9 ± 3.3ng/L TSAT 14.4 ± 2.9%		Hb 80.1 ± 9.9 g/L;SF 14.9 ± 4.1ng/L TSAT 15.8 ± 2.8%
XiuLing Wu [14]	Unknow	Hb 119.97 ± 6.24 g/L, SF 292.26 ± 1363.43 ug/L TSAT 28.27 ± 10.03%	Hb 118.40 ± 5.89 g/L, SF 225.6(170–308) ug/L TSAT 29.9 ± 6.60%	Hb118.17 ± 6.55 g/L;SF 309.39 ± 112.16ug/L TSAT 26.63 ± 9.28%	Hb119.21 ± 5.79 g/L;SF 310.75(190.18–427.95)ug/L TSAT 31.15 ± 9.25%
ZhenHua Ge [13]	subcutaneous injection(Weekly,)	Hb 78.3 ± 3.01 g/L;SF & TSAT not reported		Hb 78.11 ± 3.19 g/L;SF & TSAT not reported
LuoJia Jiang [12]	iron supplementation	Hb 62.36 ± 8.21 g/L;SF & TSAT not reported		Hb 61.69 ± 8.12 g/L;SF & TSAT not reported
XiuWen Li [11]	Unknow	Hb 50.34 ± 10.34 g/L;SF & TSAT not reported		Hb 51.56 ± 9.23 g/L;SF & TSAT not reported
Yang Qiu [10]	iron supplementation	Hb 79.57 ± 6.61 g/L;SF & TSAT not reported		Hb 80.4 ± 6.2 g/L;SF & TSAT not reported
TongQing Huang [9]	Unknow	Hb 71.46 ± 8.76 g/L;SF 96.42 ± 28.44ug/L TSAT 0.15 ± 0.04%		Hb 71.24 ± 9.33 g/L;SF 94.38 ± 21.42ug/L TSAT 0.14 ± 0.09%
QiSheng Zhou [15]	iron supplementation	Hb 62.28 ± 9.35 g/L;SF & TSAT not reported		Hb 61.48 ± 9.24 g/L;SF & TSAT not reported
LaMei Xiu [16]	Unknow	Hb 64.71. ± 8.81 g/L;SF & TSAT not reported		Hb 65.14 ± 7.93 g/L;SF & TSAT not reported
DeYue Feng [17]	Oral iron(ferrous sulfate)	Hb 81.3. ± 12.7 g/L;SF & TSAT not reported		Hb 80.1. ± 13.3 g/L;SF & TSAT not reported
Ping Zhang [18]	iron supplementation	Hb 61.48. ± 7.96 g/L;SF & TSAT not reported		Hb 62.32 ± 7.28 g/L;SF & TSAT not reported
SuQin Gao [19]	iron supplementation(100-200 mg)	Hb 67.05. ± 9.37 g/L;SF & TSAT not reported		Hb 66.78 ± 9 0.21 g/L;SF & TSAT not reported
Jun Wu [20]	Oral iron(100-200 mg)	Hb 71.95 ± 14.32 g/L;SF & TSAT not reported		Hb 67.05 ± 9.37 g/L;SF & TSAT not reported
Pablo E. Pergola [21]	Oral iron(TIW,QW,Q2W)	Hb 9.71 ± 0.74 g/L;SF 1.59 ± 1.0Pmol/L TSAT not reported	Hb 9.81 ± 0.745 g/L;SF 1.27 ± 0.91Pmol/L TSAT not reported	Hb 9.63 ± 0.86 g/L;SF 1.81 ± 0.91Pmol/L TSAT not reported
Young Ki Lee [22]	subcutaneous injection	Hb 10.7 ± 0.8 g/dL;SF 338 ± 424 ng/mL TSAT 29.8 ± 12.8%		Hb 10.5 ± 0.7 g/dL;SF 306 ± 324 ng/mL TSAT 35.4 ± 16.4%
Gabriel Mircescu [23]	subcutaneous injection(Weekly,2weekly)	Hb 11.4 ± 1.0 g/dL;SF 333(248–460)ng/L TSAT 0.15 ± 0.04%		Hb 11.4 ± 0.8 g/dL; SF 339(254–391)ng/L TSAT 0.15 ± 0.04%
Lars G. Weiss [24]	Pre-study: If serum ferritin (SF) was < 200 µg/L, a loading dose of 100 mg iron sucrose was given at each of five consecutive dialysis sessions (total 500 mg). Maintenance: If SF remained < 200 µg/L, weekly 50 mg or alternate-week 100 mg iron sucrose was continued until SF was again ≥ 200 µg/L.	Hb 11.1 ± 1.2 g/dL;SF 490 ± 218ug/L TSAT 31 ± 18%		Hb 11.1 ± 1.2 g/dL;SF 540 ± 319ug/L TSAT 26 ± 12%
Jian Xiong [25]	iron supplementation	Hb 111.14 ± 14.45 g/L;SF 152.08 ± 124.10ug/L TSAT not reported		Hb 118.07 ± 11.33 g/L;SF 107.09 ± 140.99ug /L TSAT not reported
Nan Chen [26]	unknow	Hb 8.3 ± 0.8 g/dL;SF 445 ± 456.5 ng/mL TSAT 25.5 ± 11.1%		Hb 8.2 ± 0.1 g/dL;SF 321 0.9± 232.2 ng/mL TSAT 23.9 ± 6.2%
Mehta, K S [27]	subcutaneous injection	Hb 8.81 ± 0.76 g/dL;SF & TSAT not reported		Hb 8.86 ± 0.89 g/dL;SF & TSAT not reported
Shubhadeep D. Sinha [28]	unknow	Hb 8.39 ± 0.90 g/dL;SF & TSAT not reported		Hb 8.80 ± 0.89 g/dL;SF & TSAT not reported
Wen-Yi Li [29]	subcutaneous injection	Hb 10.46 ± 0.22 g/dL;SF 347 ± 118 ng/mL TSAT not reported		Hb 9.78 ± 0.21 g/dL;SF & TSAT not reported
Michael Allon [30]	unknow	Hb 11.0 ± 0.5 g/dL; TSAT 29(20–47)% SF not reported	Hb 11.0 ± 0.6 g/dL; TSAT 31(20–59)% SF not reported	Hb 11.0 ± 0.7 g/dL; TSAT 26(20–47)% SF not reported

However, beyond conventional patient-level confounders—such as baseline hemoglobin, iron indices, and inflammatory status—considerable heterogeneity persists in the “how” of administration: dosing frequency (once versus thrice weekly), titration logic (fixed-dose versus response-driven), and injection technique (needle length, site rotation, adherence support). These application-level factors are intricately linked with the route of administration: subcutaneous cohorts typically utilize low-frequency, high-volume regimens, whereas intravenous protocols favor high-frequency, low-dose schedules, amplifying a three-way interaction between route, frequency, and dose. Unless this bundle is disentangled, the apparent superiority of subcutaneous delivery may simply reflect optimized application patterns rather than the route itself. Therefore, there is an urgent need to systematically evaluate the independent impact of the “route of administration” on therapeutic outcomes to clarify how much of the patient’s response is genuinely determined by subcutaneous versus intravenous delivery—rather than being confounded by concomitant differences in frequency, dose, or technique.

Iron supplementation is a standard approach for correcting iron-deficiency anemia; however, its effect on hemoglobin (Hb) concentration depends not only on the dose and route of administration but also on non-pharmacological factors such as baseline Hb level, iron metabolic status, and individual patient characteristics. Qualitative analyses reveal that, although oral iron is the most common formulation, its capacity to raise Hb is modest: in most trials, the mean Hb difference between iron-treated and control groups is less than 3 g/L, and efficacy is further diminished in patients with chronic inflammation or systemic disease. Intravenous or subcutaneous iron repletes iron stores more rapidly but confers no clear superiority in improving Hb; rather, its principal benefit is maintaining Hb stability, particularly in dialysis-dependent or peri-operative cohorts. Notably, baseline Hb is a critical determinant of response: individuals with severe anemia (Hb < 90 g/L) seldom achieve meaningful increments despite moderate to high iron doses, underscoring the need for adjunctive therapies such as erythropoiesis-stimulating agents. Serum ferritin (SF) and transferrin saturation (TSAT) also modulate outcomes; functional iron deficiency (high SF, low TSAT) markedly attenuates the therapeutic effect. In summary, while iron therapy is fundamental to anemia management, its impact on Hb is predominantly stabilizing rather than substantially augmentative, and optimal use mandates the integration of baseline Hb values with comprehensive iron metabolic profiling.

Literature quality assessment

The 22 included studies [3–24] that focused on the frequency of administration and efficacy of various doses of short-acting erythropoiesis-stimulating agents (ESAs) all employed a randomized controlled design; however, none specified the method of randomization. None of the studies mentioned allocation concealment, and neither the subjects nor the researchers were blinded in any of the studies. Blind evaluation of the results was utilized in all studies. Data integrity was maintained across all studies, with no evidence of selective reporting or other biases detected.(Fig. 2).

Fig. 2.

Risk-of-bias traffic-light plot for 22 RCTs of short-acting ESAs (different dosing frequencies)of ESAs

The 6 included studies [25–30] that analyzed the efficacy of darbepoetin and high-frequency, low-dose short-acting erythropoiesis-stimulating agents (ESAs) were reviewed. Four studies [26, 27, 29, 30] reported using a randomized controlled design but did not specify the randomization method; while one study [25] employed a coin toss method for random grouping, and another study [28] used a random number table method. None of the studies mentioned allocation concealment. Additionally, neither the subjects nor the researchers were blinded in any of the studies, and blind evaluation of the results was not conducted. Data integrity was maintained across all studies, with no evidence of selective reporting or other biases detected. (Fig. 3).

Fig. 3.

Risk-of-bias traffic-light plot for 6 RCTs of darbepoetin alfa vs short-acting ESAs

Publication bias analysis

Publication bias influenced the results of the meta-analysis. This study included 22 studies on the frequency of administration and efficacy of various doses of short-acting erythropoiesis-stimulating agents (ESAs), as well as 6 studies analyzing the efficacy of darbepoetin and high-frequency, low-dose short-acting ESAs. Hemoglobin (Hb) levels were reported in all 28 studies, and a publication bias analysis was conducted using RevMan 5.4 [2]. The analysis revealed that the scatter plot distribution was asymmetrical, indicating a potential significant presence of publication bias. (Figs. 4 and 5).

Fig. 4.

Funnel plot for publication bias across 22 RCTs of short-acting ESAs (dosing frequency)

Fig. 5.

Funnel plot for publication bias across 6 RCTs of darbepoetin vs short-acting ESAs

Meta-analysis results

Effects of route of administration

Subcutaneous administration of short-acting erythropoiesis-stimulating agents (ESAs) was associated with a significantly greater increase in hemoglobin (Hb) levels (weighted mean difference [WMD] 1.49 g/dL, 95% confidence interval [CI] 0.76–2.22) compared to intravenous regimens (WMD 0.54 g/dL, 95% CI 0.41–0.67). The formal test for subgroup differences confirmed the route of administration as a significant effect modifier [31] (Chi² = 6.38, p = 0.01, I² = 84.3%) (Fig. 6).

Fig. 6.

Forest plot of haemoglobin response: subcutaneous (Experimental)vs intravenous route (Control)of short-acting ESAs

As this factor is not the primary focus of the present analysis, we will examine its overall effect without exploring the detailed mechanisms involved.

Effects of low-frequency high-dose short-acting ESAs versus high-frequency low-dose short-acting ESAs on hemoglobin (Hb)

Hemoglobin (Hb) was chosen as the primary indicator for evaluating clinical efficacy. The 22 studies [3–24] analyzed the frequency of administration and efficacy of short-acting erythropoiesis-stimulating agents (ESAs) at varying dose specifications, reporting hemoglobin (Hb) improvement. Heterogeneity testing revealed a p-value of less than 0.00001 and an I² value of 83%, indicating substantial heterogeneity among the studies, which prompted the use of a random-effects model. The results indicated that the difference in Hb improvement between the experimental group (low-frequency, high-dose) and the control group (high-frequency, low-dose) was not statistically significant. [22]. [MD = 1.38, 95% CI (−0.64, 3.40), Z = 1.34, p = 0.18 > 0.05] (Fig. 7).

Fig. 7.

Forest plot of low-frequency high-dose(Experimental) vs high-frequency low-dose(Control) short-acting ESAs

Although the once-weekly (qw) regimen of recombinant human erythropoietin (epoetin) delivered a higher cumulative weekly dose than either twice-weekly (biw) or thrice-weekly (tiw) schedules, it conferred no corresponding advantage in haemoglobin (Hb) response. This observation suggests that the pharmacodynamics of short-acting erythropoiesis-stimulating agents (ESAs) may reach a ceiling effect, or that more frequent dosing provides a more sustained erythropoietic stimulus, thereby stabilising Hb levels despite a lower total dose. From a pharmacoeconomic perspective, the cost per gram-per-decilitre increment in Hb is likely to be greater with qw administration, potentially imposing a heavier economic burden during long-term therapy. Nevertheless, the reduced injection frequency may enhance patient adherence and decrease healthcare utilisation. Clinical decisions should therefore integrate efficacy, cost, patient preference and individual therapeutic goals rather than be guided solely by cumulative dose. Future investigations should examine whether dosing frequency influences Hb variability, cardiovascular outcomes and health-related quality of life.

To explore the sources of heterogeneity, we conducted a pre-specified subgroup analysis by dialysis status. All trials were stratified into three categories based on dialysis status: “dialysis,” “non-dialysis,” and “mixed population.”

The dialysis subgroup (19 studies, 709 vs. 641 participants) showed a marginal trend favoring the low-frequency, high-dose regimen (mean difference [MD] = 1.62 g/L, 95% confidence interval [CI] −0.62 to 3.87); however, the confidence interval crossed zero, indicating that the difference was not statistically significant (p = 0.16). This subgroup also exhibited extremely high heterogeneity (I² = 85%, τ² = 17.67), suggesting substantial clinical variability within the dialysis population regarding baseline hemoglobin levels, dialysis adequacy, inflammatory status, and iron metabolism, which complicates straightforward pooled interpretation.

The non-dialysis subgroup comprised only two studies (140 vs. 134 participants), yielding an effect estimate of −1.03 g/L (95% CI −3.38 to 1.31) in the opposite direction; again, the difference was not significant (p = 0.39), but heterogeneity was negligible (I² = 0%, τ² = 0), indicating a stable estimate—albeit with limited statistical power.

The mixed-population category included one small trial (23 vs. 23), which yielded an MD of 1.09 g/L (95% CI −5.86 to 8.04), providing no reliable evidence.

Overall, the test for subgroup differences yielded χ² = 2.62, df = 2, p > 0.05, confirming that dialysis status did not materially modify the direction or magnitude of the intervention effect. Nevertheless, the high intra-subgroup heterogeneity within the dialysis stratum remains a critical consideration when interpreting the overall findings. Future high-quality, adequately powered randomized controlled trials (RCTs) with rigorous stratification are warranted to determine the optimal dose-frequency combination specifically for dialysis patients.(Fig. 8)

Fig. 8.

Stratified forest plot of Fig. 7 by dialysis status (dialysis, non-dialysis, mixed)

To further validate the robustness of the aforementioned subgroup findings, we conducted sensitivity analyses to evaluate the stability of the results (Fig. 9).

1. Influence analysis: After excluding the high-impact study [11], the overall I² decreased from 83% to 54%, while the pooled effect size remained clinically similar (mean difference [MD] = 0.67 g/L, 95% confidence interval [CI] −0.65 to 2.00, p = 0.32), confirming the robustness of the findings.

2. RCT-only analysis: When restricted to randomized controlled trials (n = 443 vs. 377), results from the fixed-effects model (MD = −0.95 g/L, p = 0.17, I² = 50%) and the random-effects model (MD = −0.97 g/L, p = 0.37, I² = 50%) were consistent, with confidence intervals including zero and showing the same direction. This indicates that the choice of statistical model did not have a substantial impact on the conclusions.(Figure 9)

Fig. 9.

Sensitivity analysis of Fig. 7 after excluding high-risk and open-label studies

Effects of darbepoetin versus high-frequency low-dose short-acting ESAs on hemoglobin

Hemoglobin (Hb) was selected as the primary indicator for evaluating clinical efficacy. The six studies [25–30]that analyzed the efficacy of darbepoetin compared to high-frequency, low-dose short-acting erythropoiesis-stimulating agents (ESAs) reported on Hb improvement. Heterogeneity testing indicated a p-value of 0.0008 (p < 0.05) and an I² value of 76%, suggesting substantial heterogeneity among the studies, which necessitated the use of a random-effects model. The results revealed that the difference in Hb improvement between the experimental group (darbepoetin) and the control group (high-frequency, low-dose short-acting ESAs) was not statistically significant. [32]. [MD = 0.23, 95% CI (−0.16, 0.62), Z = 1.16, p = 0.25] (Fig. 10).

Fig. 10.

Forest plot of haemoglobin response: darbepoetin alfa(Experimental) vs short-acting ESAs (Control)

To convert darbepoetin doses into epoetin equivalent units for pooled analysis, it is essential to establish a reliable conversion ratio. This strategy ensures that our analysis can accommodate uncertainties arising from different types of studies and allows for a thorough sensitivity assessment of potential variations in the conversion ratio.

In the meta-analysis comparing darbepoetin (measured in micrograms) with epoetin (measured in units), we conducted a systematic sensitivity analysis to evaluate the impact of the dose-conversion coefficient on the study conclusions. We introduced the “hemoglobin increase per unit dose” metric (calculated as hemoglobin concentration [g/dL] divided by drug dose [IU]) to quantify differences in dosing efficiency across various conversion coefficients. Since reported conversion ratios vary in the literature, this study adopted a comprehensive and robust approach by testing a range (180–300 IU/μg) that covers all reported values rather than relying on a single fixed ratio. Specifically, three key ratios—180, 200, and 300 IU/μg—were evaluated. The central point estimate of 200 IU/μg was selected, as it represents a population pharmacokinetic-based value cited in multiple studies. The lower bound of 180 IU/μg was primarily derived from pharmacokinetic (PK) studies [32], while the upper bound of 300 IU/μg was based on estimates from clinical efficacy studies. We created separate datasets for each conversion coefficient and performed meta-analyses using a random-effects model. By comparing summary effect sizes, heterogeneity statistics, and patterns of statistical significance, we assessed the influence of coefficient selection on the conclusions.

This method is methodologically sound for several reasons. First, the metric aligns with the fundamental principle of pharmacodynamics, the “dose-response relationship,” directly linking drug dose to clinical efficacy and avoiding the limitations of comparing absolute doses alone. Second, systematically varying the conversion coefficient allows for a comprehensive assessment of the sensitivity of the conclusions to coefficient choice. Finally, this approach provides a unified benchmark for efficiency comparison across all included studies. Although the standard deviation estimate is based on the coefficient of variation rather than raw individual data, the sensitivity analysis focuses on trends in effects rather than absolute values. Moreover, multiple analyses have confirmed the robustness of the main conclusions across different methodological assumptions. This analytical framework effectively identifies the extent to which dose-conversion coefficients affect comparative results, offering a methodological reference for clinical decision-making. We will detail the specific data and analyses for different conversion coefficients in the subsequent text. To prevent loss of numerical precision, the original means (mean) and standard deviations (SD) are expressed in scientific notation (e.g., 1.64 × 10^{− 3}), retaining the units as [original unit, e.g., IU/μg]. All effect-size calculations are performed on the raw values without any artificial scaling. (Figs. 11, 12, 13)

Fig. 11.

Dose-conversion sensitivity (180 IU/μg): haemoglobin increase per unit dose

Fig. 12.

Dose-conversion sensitivity (200 IU/μg): haemoglobin increase per unit dose

Fig. 13.

Dose-conversion sensitivity (300 IU/μg): haemoglobin increase per unit dose

In the six randomized controlled trials included, we compared the erythropoietic efficacy of darbepoetin with that of short-acting epoetin. Across the studies, darbepoetin consistently demonstrated a greater improvement in hemoglobin levels (Figs. 1–3. Despite substantial heterogeneity (I² = 100%), the random-effects pooled estimates were as follows:

Fig. 1 Mean difference −0.62 × 10^{− 3} (95% CI −1.25 to 0.01) × 10^{− 3}, p = 0.06, indicating a borderline significant trend favoring darbepoetin;

Fig. 2 Mean difference −0.83 × 10^{− 3} (95% CI −1.46 to −0.19) × 10^{− 3}, p = 0.01, demonstrating a statistically significant advantage for darbepoetin

Fig. 3 Mean difference −1.45 × 10^{− 3} (95% CI −2.06 to −0.83) × 10^{− 3}, p < 0.00001, indicating a highly significant superiority of darbepoetin

Collectively, these findings suggest that, under most study conditions, darbepoetin is likely to be more effective than short-acting epoetin in achieving or maintaining target hemoglobin concentrations. Although this study did not directly assess dose equivalence between the two agents, all included trials employed hemoglobin-targeted dose-titration strategies; thus, the observed differences in efficacy are a better reflection of real-world clinical benefit. Admittedly, conversion factors such as 200 IU per microgram may influence individual patient responses; however, since our analysis focused on clinical endpoints (i.e., change in Hb concentration), the findings remain clinically relevant. In addition, the substantial heterogeneity across studies (I² = 100%) indicates variability in patient populations, baseline Hb levels, dosing frequencies, and adjustment protocols. Future investigations should standardize treatment regimens to allow for a more precise comparison of the relative effectiveness of the two preparations

Our results showed no significant difference in hemoglobin (Hb) response across different dosing frequencies, seemingly indicating equivalent clinical efficacy between long-acting and short-acting erythropoiesis-stimulating agents (ESAs). It should be emphasized, however, that this “absence of difference” refers only to the final Hb level and its stability, not to the intrinsic pharmacodynamic efficiency of the drugs. In dose-conversion analyses, darbepoetin achieved a comparable Hb response while requiring significantly fewer cumulative international units (IU) than epoetin alfa, implying a higher unit potency. This observation can be explained by darbepoetin’s longer half-life and flatter concentration–effect curve: despite less frequent administration, the integrated drug exposure remains sufficient to maintain stable Hb with a lower total exposure. Thus, “equivalent efficacy” and “superior dosing” are not contradictory; the former reflects the clinical endpoint, whereas the latter reflects pharmacodynamic efficiency. Together, they support darbepoetin’s advantages in reducing injection burden and improving dose economy.

Comparison of safety between low-frequency high-dose short-acting ESAs and high-frequency low-dose short-acting ESAs in the treatment of renal anemia

The 18 studies [3–10, 12, 13, 15–22] that reported safety outcomes for the analysis of efficacy and frequency of different dose specifications of short-acting erythropoiesis-stimulating agents (ESAs) were included. Heterogeneity testing revealed a p-value of 0.78 (p > 0.05) and an I² value of 0%, indicating no significant heterogeneity among the studies, which warranted the use of a fixed-effects model. The results indicated that the difference in safety between the experimental group (low-frequency, high-dose) and the control group (high-frequency, low-dose) was not statistically significant. [MD = 1.30, 95% CI (0.98, 1.73), Z = 1.82, p = 0.07] (Fig. 14).

Fig. 14.

Safety comparison: low-frequency high-dose(Experimental) vs high-frequency low-dose short-acting ESAs(Control)

Comparison of safety between Darbepoetin and high-frequency low-dose short-acting ESAs in the treatment of renal anemia

The six studies [25–30] that analyzed safety outcomes related to the efficacy of darbepoetin compared to high-frequency, low-dose short-acting erythropoiesis-stimulating agents (ESAs) were included. Heterogeneity testing indicated a p-value of 0.90 and an I² value of 0%, suggesting no significant heterogeneity among the studies, which justified the use of a fixed-effects model. The results revealed a statistically significant difference in safety between the experimental group (darbepoetin) and the control group (high-frequency, low-dose), with darbepoetin demonstrating superior safety for the treatment of renal anemia. [MD = 0.63, 95% CI (0.41, 0.97), Z = 2.09, p = 0.04] (Fig. 15).

Fig. 15.

Safety comparison: darbepoetin alfa vs short-acting ESAs

Discussion

This systematic review and network meta-analysis compared the efficacy and safety of distinct erythropoiesis-stimulating agent (ESA) regimens for anemia complicating chronic kidney disease (CKD), focusing on two key questions: (i) whether a low-frequency, high-dose short-acting ESA strategy is non-inferior to the traditional high-frequency, low-dose approach, and (ii) whether long-acting darbepoetin alfa offers incremental value over short-acting preparations.

Our findings indicate that, from a hemoglobin-centric perspective, extending the dosing interval of short-acting ESAs to once weekly while proportionally increasing the single dose yields an equivalent increment in Hb concentration (pooled mean difference [MD] ≈ 1 g L^{− 1}; 95% confidence interval [CI] crossing null). This observation, largely consistent across dialysis, non-dialysis, and mixed populations, supports the adoption of a weekly subcutaneous schedule as a valid alternative that preserves efficacy while markedly improving patient adherence and reducing injection-site burden.

Although the cumulative weekly dose of epoetin was higher with the once-weekly (qw) regimen than with the twice-weekly (biw) or thrice-weekly (tiw) schedules, no corresponding advantage in hemoglobin (Hb) response was observed. This finding suggests that the pharmacodynamics of short-acting erythropoiesis-stimulating agents (ESAs) may reach a ceiling effect, or that more frequent dosing maintains a more continuous erythropoietic stimulus, thereby stabilizing Hb levels despite a lower total dose. From a pharmacoeconomic perspective, the cost per unit increase in Hb is likely to be greater with qw administration, potentially imposing a higher economic burden during long-term therapy. Nevertheless, the reduced injection frequency may enhance patient adherence and decrease healthcare utilization. Therefore, clinical decisions should integrate efficacy, cost, patient preference, and individual therapeutic goals rather than be guided solely by cumulative dose. Future studies should examine whether dosing frequency influences Hb variability, cardiovascular outcomes, and health-related quality of life.

When the comparator class is shifted to long-acting erythropoiesis-stimulating agents (ESAs), darbepoetin alfa achieves similar hemoglobin (Hb) targets but with a statistically and clinically lower incidence of composite adverse events (relative risk [RR] 0.63, 95% confidence interval [CI] 0.41–0.97). Coupled with a three-fold longer terminal half-life and the feasibility of monthly or bi-weekly administration, darbepoetin confers advantages in maintaining Hb stability, minimizing nursing time, and potentially mitigating immunogenic exposure—benefits that are particularly relevant for patients undergoing lifelong ESA therapy.

Caveats are warranted, however. “Efficacy equivalence” should not be equated with “therapeutic parity.” Hemoglobin response is only one domain of the decision matrix; divergences in cost-effectiveness, health-related quality of life, treatment satisfaction, and long-term vascular outcomes remain inadequately explored. Moreover, the aggregate methodological quality of the evidence base is modest: the majority of trials were open-label, lacked registered protocols, and employed heterogeneous definitions for safety events, thereby limiting granular inference on harm profiles.

A further source of uncertainty lies in the entanglement of route, frequency, and iron supplementation. Subcutaneous administration, almost universally adopted by low-frequency arms, independently augmented Hb by 1.5 g L^{− 1}; consequently, the observed non-inferiority of weekly regimens may partly reflect route-related pharmacokinetic superiority rather than scheduling per se. Similarly, divergent iron repletion strategies (intravenous vs. oral vs. none) were neither standardized nor consistently reported, introducing residual confounding that cannot be fully adjusted through meta-regression.

Future investigations should prioritize large, double-blind, stratified randomized controlled trials (RCTs) that rigorously control for administration route, iron status, and dialysis modality, while incorporating patient-reported outcomes and cost-utility endpoints. Until such data emerge, clinicians can reasonably adopt low-frequency subcutaneous erythropoiesis-stimulating agents (ESAs)—preferably darbepoetin alfa—as the default strategy for CKD-associated anemia, individualizing dose and interval to achieve the lowest effective exposure that maintains hemoglobin levels within the target corridor of 100–120 g L^{− 1}.

In summary, current evidence supports a shift from “one-size-fits-all” frequent dosing to individualized, route-optimized, low-frequency ESA regimens, anchored by darbepoetin alfa where feasible. This paradigm promises to deliver equivalent hematological benefit with fewer injections, a lower adverse-event burden, and an improved patient experience—an efficacy-safety-convenience triad that aligns with contemporary precision nephrology goals.

Limitation

Firstly, this study relied solely on hemoglobin concentration at the beginning and end of the study as the indicator for evaluating efficacy, without assessing the ability to maintain stable hemoglobin levels or evaluating other indicators of anemia improvement. The methodological quality of the included data was relatively low, as the research designs of the included studies did not clearly specify the randomization or blinding methods. Consequently, the results of this study require further validation through high-quality research. Thus, differences in efficacy and safety cannot be ruled out, and further large-scale, well-designed randomized controlled trials (RCTs) focusing on patient-centered outcomes are necessary to evaluate the safety and efficacy of low-frequency erythropoiesis-stimulating agents (ESAs), including short-acting ESAs and darbepoetin.

Secondly, our analysis is potentially limited by heterogeneity in co-interventions across the included studies. Key factors known to influence hemoglobin response, such as the route of drug administration (intravenous vs. subcutaneous) and the use of concomitant iron supplementation (oral or parenteral), were not uniformly applied or reported. While we have presented these variables in a new table to enhance transparency and attempted a subgroup analysis for the route of administration, residual confounding remains a possibility. The observed differences in hemoglobin response between schedules may be partly attributable to imbalances in these co-interventions rather than the schedule itself. This important limitation should be considered when interpreting our findings.

At the same time, although this meta-analysis used “hemoglobin increment per unit dose” as the primary metric of erythropoiesis-stimulating agent (ESA) efficiency—allowing for a quantitative comparison across different conversion factors—it does not fully align with the “titrate-to-target” strategy that remains standard in current practice. Clinicians typically adjust the dose according to baseline hemoglobin (Hb), individual response, and complication risk until the patient reaches the target range of 100–120 g L^{− 1}, rather than aiming for the maximal Hb increment per unit dose.

Because none of the included trials provided individual patient data (IPD), our aggregate analysis may have underestimated key real-world outcomes such as treatment satisfaction, transfusion avoidance, and safety events. Moreover, the absence of patient-level titration data precluded evaluation of how conversion factors affect “time-to-target,” “dose-adjustment frequency,” and “risk of overshoot.” Prospective titration trials or analyses of real-world electronic health records will be required to validate these findings in a target-directed dosing context.

Finally, due to the heterogeneous structure of adverse-event (AE) reports in our cohort, the proportion of missing Common Terminology Criteria for Adverse Events (CTCAE) grades was high (many records lacked any severity classification), and terminology standardization as well as severity attribution varied markedly across data sources. Consequently, the conventional comparison of grade-specific AE rates was not feasible. We therefore adopted “total AE episodes per treated subject” as a surrogate safety metric. While this measure provides an overall impression of AE burden, it may dilute signals and underestimate the incidence of high-grade toxicities.

Conclusions

Despite certain limitations in this study, the findings from the 28 studies suggest that the low-frequency group is not inferior to the high-frequency group in improving hemoglobin levels in patients with renal anemia, and there is no significant difference in safety between these two groups. Several studies have indicated that darbepoetin is associated with fewer adverse events compared to short-acting, high-frequency erythropoiesis-stimulating agents (ESAs); however, further research is necessary to confirm this finding. These results provide a basis for selecting treatment regimens involving ESAs.

Abbreviations

CNKI

China National Knowledge Infrastructure

CBM

China Biology Medicine disc

RCTs

Randomized controlled trials

CKD

Chronic kidney disease

EPO

Erythropoietin

rHuEPO

Recombinant human erythropoietins

MD

Mean difference

SMD

Standardized mean difference

OR

Odds ratio

CI

Confidence interval

Hb

Hemoglobin

SF

Serum ferritin

SD

Standard deviations

Qw

Once-weekly

Biw

Twice-weekly

Tiw

Thrice-weekly

AE

Adverse-event

CTCAE

Common Terminology Criteria for Adverse Events

IV

Intravenous

PO

Per os

Author contributions

MXY and RLO were responsible for locating and selecting articles and writing-original draft preparation; XMW and ZTG conceptualized the design of the article and were responsible for supervision, writing-reviewing and editing; others reviewed and corrected the manuscript.

Funding

Weifang Key Laboratory of Integrated Traditional Chinese and Western Medicine for Chronic Renal Failure and TCM Science and Technology Development Plan of Shandong Province (M-2022265).

Data availability

The study has been registered with the Prospective Register of Systematic Reviews (PROSPERO) under ID: CRD42024596526.

Declarations

Ethics approval

This study is a meta-analysis based on previously published literature. All data were extracted from publicly available research articles. No original data collection or direct involvement of human or animal subjects was conducted, and therefore, ethical approval was not required.

Consent for publication

Not applicable.

Conflicts of interest

None.