Use of erythropoiesis-stimulating agents in children with chronic kidney disease: a systematic review

ABSTRACT Background Erythropoiesis-stimulating agents (ESAs) revolutionized the management of anaemia in chronic kidney disease (CKD) when introduced in the late 1980s. A range of ESA types, preparations and administration modalities now exist, with newer agents requiring less frequent administration. Although systematic reviews and meta-analyses have been published in adults, no systematic review has been conducted investigating ESAs in children. Methods The Preferred Reporting Items for Systematic Reviews and Meta-analyses statement for the conduct of systematic reviews was used. All available literature on outcomes relating to ESAs in children with CKD was sought. A search of the MEDLINE, CINAHL and Embase databases was conducted by two independent reviewers. Inclusion criteria were published trials in English, children with chronic and end-stage kidney disease and use of any ESA studied against any outcome measure. An assessment of risk of bias was carried out in all included randomized trials using the criteria from the Cochrane Handbook for Systematic Reviews of Interventions. Two tables were used for data extraction for randomized and observational studies. Study type, participants, inclusion criteria, case characteristics, follow-up duration, ESA type and dosage, interventions and outcomes were extracted by one author. Results Of 965 identified articles, 58 were included covering 54 cohorts. Six were randomized trials and 48 were observational studies. A total of 38 studies assessed the efficacy of recombinant human erythropoietin (rHuEPO), 11 of darbepoetin alpha (DA) and 3 of continuous erythropoietin receptor activator (CERA), with 6 studies appraising secondary outcome measures exclusively. Recruitment to studies was a consistent challenge. The most common adverse effect was hypertension, although confounding effects often limited direct correlation. Two large cohort studies demonstrated a greater hazard of death independently associated with high ESA dose. Secondary outcome measures included quality of life measures, growth and nutrition, exercise capacity, injection site pain, cardiovascular function, intelligent quotient, evoked potentials and platelet function. Conclusions All ESA preparations and modes of administration were efficacious, with evidence of harm at higher doses. Evidence supports individualizing treatments, with strong consideration given to alternate treatments in patients who appear resistant to ESA therapy. Further research should focus on randomized trials comparing the efficacy of different preparations, treatment options in apparently ESA-resistant cohorts and clarification of meaningful secondary outcomes to consolidate patient-relevant indices.


INTRODUCTION
Chronic kidney disease (CKD) is a substantial global health burden, with mortality rates for children with end-stage kidney disease (ESKD) 55 times higher than the general paediatric population [1]. Anaemia is a common complication observed in up to 73% of children with CKD stage 3 and 93% in stages 4 and 5 [2,3].
The primary cause of this anaemia is a deficiency of erythropoietin (EPO). EPO is a 30.4-kDa glycoprotein that stimulates red cell production, differentiation and survival [4]. EPO gene expression is upregulated by hypoxia-inducible transcription factor (HIF), although in CKD the response to hypoxia is deranged, resulting in impaired production and reduced HIF-binding capacity [5][6][7].
The short half-life of rHuEPO necessitates administration three times per week [10]. In the late 1990s, darbepoetin alpha (DA) was synthesized through 'glycoengineering' amino acid changes to rHuEPO, extending its half-life to allow once-or twice-weekly dosing [11]. In 2007, continuous erythropoietin receptor activator (CERA) usage was approved, with the addition of a methoxy-polyethylene glycol polymer further prolonging the half-life to permit fortnightly or monthly dosing [12].
In adults, ESA therapy is associated with hypertension, stroke, vascular access thrombosis and overall mortality when higher haemoglobin (Hb) levels (>12.5 g/dL) are targeted [13,14]. In children this association is less clear-one large retrospective cohort study of 1569 children found no relationship [15]. The Kidney Disease: Improving Global Outcomes (KDIGO) 2012 guidelines recommend modest Hb targets of 11.0-12.0 g/dL with initial doses of 60-150 IU/kg/week for rHuEPO and 0.45 μg/kg/week for DA [16]. There also appears to be an independent association with mortality when high ESA doses are administered [17,18], therefore KDIGO specifically cautions against dose escalation in failed responders [16].
This systematic review will appraise studies assessing the efficacy of ESAs in children with CKD. It will also appraise the extent to which a safety profile has been established, while outlining all other secondary outcomes explored.

METHODS
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement for the conduct of systematic reviews was used. of 28 observational studies evaluated efficacy, of which 22 evaluated subcutaneous or intravenous administration. All of these confirmed improvements in indices of anaemia with rHuEPO administration.
The first paediatric observational study in 1989 highlighted that Hb could be successfully maintained in five paediatric dialysis patients on subcutaneous treatment, reducing the requirement for transfusion and subsequent development of anti-human leucocyte antigen (HLA) antibodies [31]. At 450 U/kg/week, the dose used was three times the upper limit of current KDIGO recommendations, and three of the five patients developed worsening hypertension.
Two studies used fixed dose regimens, with the remainder titrating dosing [31,32]. Initial doses ranged from 30 to 450 U/kg/week, with target haematocrit (HCT) ranging from 0.33 to 0.40 L/L and Hb from 9 to 13 g/dL. Dose frequency was usually three per week, although two studies explored weekly dosing. Goldraich and Goldraich [32] demonstrated the efficacy of once weekly 150 U/kg dosing in six children on continuous ambulatory PD (CAPD). Ongkingco et al. [33] found no significant decrease in HCT after 8 weeks, decreasing from thrice to once weekly maintenance dosing (with associated cost-benefit), although the study suffered from significant dropouts resulting in only seven recruits. The majority of observational studies investigated small cohorts of between 5 and 24 children (mean 15), although two larger multicentre studies were conducted in 1991 and 1994 [34][35][36]. The earlier of these included 120 children across multiple European centres [35,36], reporting a mean final dose requirement of 138 U/kg/week. The second recruited 115 ESA-naïve children treated with rHuEPO for up to 1 year [34]. A total of 81% achieved a target Hb of 9.6-11.2 g/dL, although 68% of 'nonresponders' were transplanted earlier. The median maintenance dose for children <30 kg was 225 U/kg/week and 107 U/kg/week for children >30 kg.
Six observational studies investigated intraperitoneal administration [37][38][39][40][41][42]. The first trial by Offner et al. [37] was halted early due to a high rate of peritonitis. Subsequently, Reddingius et al. [41] trained parents to inject rHuEPO into overnight 20 mL/kg bags, demonstrating a reduced requirement for transfusion without an increased peritonitis incidence. Reddingius et al. [42] and Kausz et al. [38] demonstrated in small cohorts of 10 and 14 patients, respectively, that intraperitoneal administration could maintain Hb when switched from subcutaneous rHuEPO without a significant dose increase. Administration was via a 50-mL intraperitoneal daytime dwell and Reddingius et al. [42] also demonstrated a mean dose reduction with this method against a 250-mL prolonged dwell (266 → 234 U/kg/week). Kausz  The largest study was conducted by Rusthoven et al. [39], who followed 20 ESA-naïve children for up to 1 year after starting rHuEPO in three divided doses delivered in 50-mL bags. They were able to maintain target Hb levels with a modest dose of 179 U/kg/week and with a low peritonitis incidence of 1 per 11.2 patient-months.
Three studies were randomized trials [30,[43][44][45]. Morris et al. [44] undertook a single-blinded placebo-controlled randomized crossover trial in 11 ESA-naïve children, demonstrating a significant increase in the median Hb from 7.3 to 11.2 g/dL (P < .001). Yalçınkaya et al. [43] randomized 20 ESAnaïve children on CAPD to receive low-(50 U/kg/week) or highdose (150 U/kg/week) rHuEPO for 6 months and found that while both doses were efficacious, the higher dose led to a statistically significant increase in the mean arterial BP from 85 to 101 mmHg. Four participants in the high-dose arm had to temporarily discontinue therapy due to uncontrolled hypertension, with two instances of hypertensive encephalopathy. Brandt et al. [30] randomized 44 children to low (150 U/kg/week) and high (450 U/kg/week) dosing for 12 weeks or until a 10 g/dL target Hb was reached. Attainment of the Hb target in the higher dose cohort was more rapid, though with a non-significant higher incidence of hypertension [high dose 38%, low dose 21% (P = .17)].  A further three studies examined secondary outcomes only and are outlined below [18,46,47].

DA.
A total of 11 studies investigated DA efficacy in 411 children. There were two randomized trials and nine observational studies (five prospective case series, one retrospective case series, one pro-and retrospective case series, one prospective case-control, one retrospective case-control). Two included children on PD, one included children on HD, three included conservatively managed CKD and three were mixed. Two analyzed DA in ESA-naïve children, three included children established on an ESA and the remaining four included a mixture of naïve and ESA-treated children. All demonstrated that DA was efficacious in reaching a specified Hb target. Targets were varied and generally aimed for 11-13 g/dL, although only two studies matched their target to the KDIGO recommendation of 11-12 g/dL [48,49]. Cohorts within the observational studies varied between 3 and 39 (mean 19) participants.
Dosing regimens and adjustment strategies varied in the observational studies. Initial dosing was reported between 0.27 and 1.59 μg/kg/week, with both weekly and fortnightly dosing trialled, although most starting doses were close to the KDIGO recommendation of 0.45 μg/kg/week. All studies titrated dosing.
The first observational study, conducted by De Palo et al. [50], recruited seven children titrated to intravenous DA from rHuEPO using a conversion factor (weekly epoetin alfa dose/200 = weekly DA dose). An initial mean dose of 1.59 ± 1.19 μg/kg coincided with two cases of hypertension with a rapid increase in Hb to >13 g/dL, necessitating intermittent discontinuation of treatment. The mean dosage at 3 months was 0.51 ± 0.18 μg/kg/week and the authors subsequently recommended a long-term dose of 0.25-0.75 μg/kg/week.
In a French multicentre study of 39 children, Andre et al. [51] reported an almost 2-fold higher mean dose requirement in patients switched to DA from rHuEPO as compared with ESA-naïve children [0.73 versus 0.34 μg/kg/week (P = .015)]. This was not replicated in other studies involving both ESA-naïve children and children on rHuEPO [52][53][54].
A prospective case-control study compared the efficacy of rHuEPO to DA [48]. Can et al. [48] split 34 children equally to receive rHuEPO 2-3/week or DA weekly and found no differences in the efficacy or adverse effects profile between either group.
Durkan et al. [55] and Libudzic-Nowak et al. [56] specifically investigated infants <1 year of age. Durkan et al. [55] found that only 50% of the six patients recruited reached target Hb levels of 10-11 g/dL despite a high mean administered dose of 1.2 μg/kg/week and normal iron studies. Libudzic-Nowak et al. [56] achieved target Hb concentrations of 10.7-12 g/dL in three infants ages 1, 4 and 7 months, but requiring doses of 0.3-0.7 μg/kg/week, generally higher than in older children.
One retrospective case-control study appraised intraperitoneal administration. Rijk et al. [54] evaluated 19 children, 8 of whom were previously on intraperitoneal rHuEPO. A high median dose of 0.79 μg/kg/week was required to sustain Hb levels at a mean of 11.5 ± 1.2 g/dL. Six cases dropped out due to transplantation, with a relatively low peritonitis incidence of one episode every 25.1 months.
Two randomized trials investigated DA efficacy [28,29]. Warady et al. [28] conducted an open-label non-inferiority trial in 124 children randomized (1:2) to ongoing rHuEPO therapy or DA, with results demonstrating an equivalent mean change in Hb over 28 weeks. The same team performed a prospec-tive, multicentre double-blind randomized controlled trial of 114 ESA-naïve children comparing weekly versus fortnightly titrated dosing [29]. This showed that the mean time to target Hb of 10-12 g/dL was equivalent (22 days and 24 days, respectively), although a greater proportion of patients on weekly dosing reached the target Hb at 24 weeks (98% versus 84%).
A further three studies evaluated secondary outcomes only and are discussed below [57][58][59].

CERA.
No randomized trials were identified regarding CERA use in children. Three observational studies evaluated CERA in 92 children [49,60,61]. Cano et al. [61] studied 16 children over 6 months converted from rHuEPO to fortnightly subcutaneous CERA. They found Hb was maintained, although dosing varied significantly (0.5-2.9 μg/kg/dose). Wedekin et al. [49] conducted a prospective case series on 12 children after renal transplant using a monthly intravenous dosing regimen. After 6 months of follow-up, they demonstrated an increase in mean Hb in ESAnaïve patients and maintained Hb levels in patients switched from DA (although only 75% achieved a target of 11-12 g/dL). Fischbach et al. [60] conducted an open-label multicentre study on 64 children on stable ESA regimens. An intermediate conversion factor (4 mg every 4 weeks for each weekly dose of 250 IU epoetin alfa/beta or 1.1 mg DA) derived from adult studies was tested against a twice higher conversion factor over 40 weeks. The intermediate factor proved less adequate at maintaining stable Hb, with mean Hb dropping below the lower target threshold of 10 g/dL on several occasions, whereas the higher factor was associated with more stable target Hb levels.

Secondary outcome measures
Safety. Most observational studies included a discussion of adverse effects, the most common being hypertension. Three studies specifically focussed on safety in large cohorts [18,62,63].
Borzych-Duzalka et al. [63] prospectively appraised the anaemia management of 1394 children on PD across 30 countries between 2007 and 2011 for up to 48 months. Of 1147 patients where the ESA dose was available, 2.1% with lower dose regimens (<6000 IU/m 2 /week) versus 5.3% with higher dose regimens (not specified) died (P = .02). Regression analysis demonstrated an independent increased risk of death on PD with higher ESA doses [hazard ratio (HR) per 1000 IU/m 2 /week 1.33; P < .01]. Children were more likely to be ESA sensitive with higher albumin levels, low serum parathyroid hormone and persisting diuresis.
Lestz et al. [18] conducted a retrospective cohort study using 12-to 18-month follow-up of mortality records linked to a US 2005 ESKD registry in 820 children on dialysis and ESA therapy who had not undergone transplantation during 12-18 months of follow-up. Over the observation period, 60 children (7%) died, primarily attributed to cardiovascular causes. ESAs were prescribed to 95% of survivors and 93% of those who died. Average ESA doses were significantly higher in those who died versus survivors [rHuEPO 502 versus 290 units/kg/week (P < .001), DA 0.59 versus 2.6 μg/kg/week (P < .001)] and multivariate analysis demonstrated an HR of death of 3.37 in a high-dose group (EPO ≥350 units/kg/week or DA ≥1.5 μg/kg/week) when compared with a lower reference range (EPO 100-<200 units/kg/week or DA 0.49-1.0 μg/kg/week). This finding was independent of a wide range of factors, including cause of ESKD, dialysis modality, access and achievement of a minimum target Hb level of 11 g/dL. Schaefer et al. [62] conducted an observational registry study of 319 children across 37 centres, the most comprehensive study of the safety of DA in children. Children were followed for up to 2 years, although 176 children withdrew earlier. A total of 162 patients, 50.8% of the cohort, reported a total of 434 serious adverse events (SAEs), the most common of which were peritonitis (n = 32), gastroenteritis (n = 19) and hypertension (n = 13). The authors state that this is comparable with a general cohort of children with CKD.
Four patients (1.3%) suffered six documented serious adverse drug reactions (SADRs): arteriovenous fistula thrombosis, priapism, thrombocytopenia, haemolysis, haemolytic anaemia and partial blindness. The authors suggest the latter four SADRs had more plausible explanations than related to ESA administration. Six fatal adverse events occurred, but none were considered to be related to ESA administration. No new safety issues were identified.
Two studies primarily focussed on efficacy also included safety extensions to their trials. Warady et al.'s [28] open-label non-inferiority study of DA versus rHuEPO included documentation of adverse events deemed by the investigator to be treatment related, affecting 14% (n = 6) of the rHuEPO cohort and 20% (n = 16) of the DA cohort. Injection site pain was the most common adverse event [12% (n = 5) rHuEPO, 11% (n = 9) DA], with hypertension in three of the DA cohort, one instance of vascular access thrombosis in both cohorts and access stenosis in one in the DA cohort.
Fischbach et al. [57] included a 1-year safety extension to their trial of CERA, including 37 children. It found no additional safety signals, with two SAEs, both vascular access thromboses. Hypertension was reported as an adverse event in 13% [60].

Quality of life (QoL).
Three studies assessed QoL [29,34,45]. Two studies used a non-validated self-designed questionnaire in children on rHuEPO. Small patient numbers for analysis in the first study (n = 7) prevented meaningful conclusions, while the second lacked any control arm but did demonstrate an improved QoL from baseline [34,45]. Warady et al. [29] used the Pediatric Quality of Life Inventory (PedsQL) score to assess changes in QoL in their RCT cohort of 114 children starting DA. The authors noted a statistically significant increase in the PedsQL score from baseline to 6 months (QW: 61.1 → 68.1, Q2W: 62.6 → 67.2).

Growth and nutrition.
Five papers studied aspects of growth. Scigalla et al. [35] employed a height score (see Figure 3), finding no significant changes. Two studies assessed small cohorts of six participants [64][65][66]. Rees et al. [66], analysing Rigden et al.'s [64] 1990 cohort of six children on HD, described small improvements in growth velocity in the three youngest children over 1 year, with no appreciable effect in older participants. Scharer et al. [65] noted improvements in height standard deviation scores in the two youngest children in their cohort over ∼1 year of rHuEPO therapy [−1.8 → −1.0 and −3.7 → −2.5 standard deviation score], with minimal changes in four older children.
Stefanidis et al. [67] found no significant change in growth in 10 children 1 year after anaemia correction. These papers were summarized as the subject of a 1996 review [68]. Subsequently, Boehm et al. [46] conducted a retrospective cohort study in 47 children followed from initial referral to pre-dialysis care and after the initiation of dialysis. They reported that rHuEPO therapy initiation at referral was the only modifiable factor independently associated with a catch-up growth velocity once dialysis was initiated {odds ratio (OR) 6.67 [95% confidence interval (CI) 1.00-44.10], P < .05}.
Exercise capacity. Five studies investigated exercise capacity using treadmill tests (see Table 4). Baraldi et al. [69] demonstrated improvements in several domains using an unspecified treadmill protocol 2-4 weeks following anaemia correction in seven children. Rigden et al. [64] demonstrated an improved treadmill time using the modified Bruce protocol in four children on HD. Warady et al. [70] assessed nine children undergoing PD, demonstrating improvements in all parameters using the Balke protocol 1 month following achievement of the target HCT of 30%. It was the only study that used controls, comparing results with five age-matched children without renal disease and confirming a significant improvement in children with renal disease. Martin et al. [71] found mild sustained improvements in treadmill time in 12 children. Morris et al. [45] included exercise testing, although the results were unpublished. Other secondary outcome measures. Infrequently considered outcome measures included intelligence quotient (n = 1) [73], platelet function (n = 2) [71,73] and evoked potentials (n = 1) [71,74].

DISCUSSION
ESAs have transformed the management of renal anaemia, reducing transfusion burden and HLA sensitization. They are widely used in the USA and European Union, where up to 94% of children on HD are prescribed a regular ESA [76,77]. Yet challenges remain-the European Dialysis Transplant Association registry reported in 2012 that 33.4% of children on dialysis <2 years of age and 31.2% >2 years had Hb levels below target [77].
This systematic review identifies a highly heterogeneous collection of studies assessing the use of ESAs in children. The challenges of recruiting within a paediatric cohort were apparent, with larger datasets requiring the involvement of multiple centres across countries.
Early studies of rHuEPO were characterized by small prospective observational cohorts demonstrating efficacy whether given subcutaneously or intravenously, while identifying that higher doses were associated with adverse events such as hypertension and vascular thrombosis. ESAs were shown to be less effective in the presence of iron deficiency and most subsequent studies ensured adequate iron stores.

Table 4. Assessments of exercise capacity
Martin et al. [71] Warady et al. [70] Baraldi et al. [69] Rigden et al. [ A number of other secondary measures were explored using rHuEPO, varying from patient-relevant measures such as exercise tolerance and quality of life to physiological parameters, including cardiac function, evoked potentials, growth and nutrition, and platelet function. These were conducted on small cohorts.
The randomized placebo-controlled crossover trial and casecontrol studies conducted by Morris et al. [44,47] suggest improvements in cardiac function following anaemia correction. Transplant recipients established on ESAs demonstrate comparably more minor cardiovascular improvements following transplantation when compared with CKD patients. This suggests that anaemia rather than uraemia correction plays a greater role in improving cardiac health or that other factors may be more important post-transplantation [47].
Studies on DA generally featured larger cohorts demonstrating non-inferiority against rHuEPO and established a similar safety profile [28,62]. Weekly and fortnightly dosing both appear feasible treatment options [29]. QoL was explored in one study [29]. A modest increase in the PedsQL score was noted after 6 months of treatment, a finding supported by larger crosssectional studies that demonstrate improved QoL in children with CKD without anaemia compared with those with persistent anaemia [78]. Further interrogation of outcomes relevant to patients has not been forthcoming. This review concurs with a Cochrane review of 2014 that noted that 'formulations based on patient centred outcomes … are sparse and poorly reported' [22]. More studies incorporating patient-centred outcomes are required to strengthen the rationale for intervention and choice of agent.
Early studies on CERA demonstrated efficacy in small paediatric cohorts and that Hb could be maintained when switching from other ESA preparations [60]. A higher conversion factor than that used in adults when changing from other ESA preparations may be required, and the safety profile appears similar to other ESAs [18]. Randomized trials comparing dosing regimens, comparing CERA with other ESAs and comparing patient preferences are lacking. A further dosing study is currently under way [79].
Intraperitoneal administration was predominantly evaluated using rHuEPO. It appears feasible and safe and is supported by pharmacokinetic studies demonstrating comparable bioavailability to other routes [80][81][82]. Nevertheless, it remains an uncommon route of administration. Intraperitoneal DA appears non-inferior to intraperitoneal rHuEPO, although only one study was identified.
Small studies on infants demonstrate that particularly high ESA doses may be required [55,56]. Larger observational studies have also demonstrated higher ESA dose requirements in younger cohorts that appear consistent across ESA types [16,34,63]. One suggested reason for this apparent ESA resistance is a greater prevalence of iron deficiency: a study of anaemia in 2899 children on dialysis enrolled in the United States Renal Data System between 1996 and 2000 found that children ages 0-4 years were least likely to achieve target Hb, correlating with the lowest use of intravenous iron (33.9% versus 71%, ages [15][16][17][18][19] [83]. In contrast to this, Borzych-Duzalka et al.'s [63] study of 1394 children enrolled in the International Paediatric Peritoneal Dialysis Network registry between 2007 and 2011 found no relationship between Hb levels and iron supplementation, with an inverse association between Hb and ferritin levels (although transferrin saturation data were not available).
This suggests other mechanisms may contribute to an apparent ESA resistance in younger children. Speculated causes include higher numbers of EPO receptors that do not contribute to erythropoiesis, potentially 'mopping up' ESAs and reducing their haematopoietic potential [84]. Borzych-Duzalka et al. [63] also found reduced dose discrepancies in younger children when weight was substituted for body surface area (BSA) as a metric, suggesting requirements may be more proportional to metabolic rate than weight-based data suggest. Further studies that compared body weight with BSA dosing may help confirm this finding. Other studies have identified markers of dialysis adequacy, indices of nutritional intake, inflammatory status and hyperparathyroidism as primary factors in determining ESA resistance rather than iron deficiency [85,86].
Nevertheless, the consistent finding of an independent relationship between higher ESA doses and mortality is of concern [18,63]. High doses of ESAs can directly cause endothelial damage, vasoconstriction and platelet activation, all of which could plausibly increase the risk of cardiovascular mortality in children [87,88]. Although the observational nature of the studies in question prevents the establishment of a definitive causal link, caution should clearly be applied when titrating ESAs in clinical practice, with careful consideration of all available interventions to maximize haemoglobin.
The most common reported adverse effect was hypertension. While some individual cases were clearly attributable to very high doses of ESAs [50,74], in general the rate of hypertension in observational studies was noted to be comparable with other CKD cohorts.
Overall, there is no evidence to recommend one ESA as more efficacious or safe than any other. Factors influencing the decision of which ESA to choose will depend on considering the most convenient means of administration, taking into account age, mode of renal replacement therapy (if any) and patient preference. The morbidity and mortality risks associated with greater dosages of ESAs mandate thorough assessment of children with apparent ESA insensitivity.