Skip to main content
. 2025 Feb 25;35(2):129–167. doi: 10.25259/ijn_389_23

Table 32:

Evidence profile for roxadustat as an alternative to ESA for DD-CKD patients

Population: Anemia in dialysis-dependent chronic kidney disease

Intervention: Roxadustat

Comparator: ESA (epoetin alpha/darbepoetin alpha)

Outcome

Time frame

Study results and measurements Absolute effect estimates

Certainty of the evidence

Plain language summary
ESA (epoetin alpha/darbepoetin alpha) Roxadustat

All-cause mortality from 6 to 52 weeks

Odds ratio: 1.11

(CI 95% 0.76–1.62)

Based on data from 1715 participants in six studies

82

per 1000

90

per 1000

Low

Due to serious risk of bias, due to serious imprecision1

We are uncertain whether roxadustat (any dose) increases all-cause mortality from 6 to 52 weeks.

Difference: 8 more per 1000

(CI 95% 18 fewer–44 more)

All-cause mortality from 108 to 209 weeks

Odds ratio: 1.13

(CI 95% 0.96–1.33)

Based on data from 3974 participants in three studies

171

per 1000

189

per 1000

Very low

Due to very serious risk of bias, due to serious imprecision2

We are uncertain whether roxadustat (any dose) increases all-cause mortality from 108 to 209 weeks.

Difference: 18 more per 1000

(CI 95% 6 fewer–44 more)

Need for iron supplementation from 6 to 52 weeks

Odds ratio: 0.57

(CI 95% 0.16–2.05)

Based on data from 1215 participants in three studies

793

per 1000

685

per 1000

Very low

Due to serious risk of bias, due to serious inconsistency, due to serious imprecision3

We are uncertain whether roxadustat (any dose) decreases the need for iron supplementation from 6 to 52 weeks.

Difference: 107 fewer per 1000

(CI 95% 413 fewer–94 more)

Need for ESA from 6 to 52 weeks

Odds ratio: 13.38

(CI 95% 0.75–238.31)

Based on data from 916 participants in two studies

0

per 1000

0

per 1000

Very low

Due to very serious risk of bias, due to very serious imprecision4

We are uncertain whether roxadustat (any dose) increases the need for ESA from 6 to 52 weeks.

Difference: 0 fewer per 1000

(CI 95% 0–0)

Need for iron supplementation from 52 to 208 weeks

Odds ratio: 0.56

(CI 95% 0.13–2.46)

Based on data from 2940 participants in two studies

288

per 1000

184

per 1000

Very low

Due to very serious risk of bias, due to serious inconsistency, due to serious imprecision5

We are uncertain whether roxadustat (any dose) decreases need for iron supplementation from 52 to 208 weeks.

Difference: 103 fewer per 1000

(CI 95% 238 fewer–211 more)

Need for ESA up to 208 weeks

Odds ratio: 20.29

(CI 95% 4.89–84.25)

Based on data from 2106 participants in one study

2

per 1000

39

per 1000

Very low

Due to very serious risk of bias, due to serious imprecision6

We are uncertain whether roxadustat increases need for ESA up to 208 weeks.

Difference: 37 more per 1000

(CI 95% 8 more–142 more)

Treatment emergent adverse events from 6 to 52 weeks

Odds ratio: 1.45

(CI 95% 1.08–1.96)

Based on data from 1715 participants in six studies

786

per 1000

841

per 1000

Moderate

Due to serious risk of bias7

Roxadustat (any dose) may increase treatment emergent adverse events from 6 to 52 weeks.

Difference: 56 more per 1000

(CI 95% 13 more–92 more)

Treatment emergent adverse events from 108 to 209 weeks

Odds ratio: 1.05

(CI 95% 0.85–1.28)

Based on data from 2935 participants in two studies

849

per 1000

855

per 1000

Very low

Due to very serious risk of bias, due to serious imprecision8

We are uncertain whether roxadustat (any dose) increases or decreases treatment emergent adverse events from 108 to 209 weeks.

Difference: 6 more per 1000

(CI 95% 22 fewer–29 more)

Patients requiring blood transfusion 6–52 weeks Odds ratio: 0.58 (CI 95% 0.42–0.82) Based on data from 821 participants in two studies

202

per 1000

128

per 1000

Very low

Due to very serious risk of bias, due to serious imprecision9

We are uncertain whether roxadustat (any dose) decreases patients requiring blood transfusion from 6 to 52 weeks.
Difference: 74 fewer per 1000 (CI 95% 106 fewer–30 fewer)
Patients requiring blood transfusion from 58 to 108 weeks

Odds ratio: 0.87

(CI 95% 0.65–1.17)

Based on data from 1869 participants in two studies

93

per 1000

82

per 1000

Very low

Due to serious risk of bias, due to serious inconsistency, due to serious imprecision10

We are uncertain whether roxadustat (any dose) decreases patients requiring blood transfusion from 58 to 108 weeks.

Difference: 11 fewer per 1000

(CI 95% 31 fewer–14 more)

Change in hemoglobin levels from baseline from 6 to 52 weeks

Measured by:

Scale: High better

Based on data from 5553 participants in nine studies

Mean Mean

Low

Due to serious risk of bias, due to serious publication bias11

We are uncertain whether roxadustat (any dose) increases change in hemoglobin levels from baseline from 6 to 52 weeks.

Difference: 0.21 lower (MD)

(CI 95% 0.11 lower–0.32 higher)

QoL assessed by EQ-5D-5L VAS

Measured by:

Scale: High better

Based on data from 783 participants in one study

Mean Mean

Very low

Due to very serious risk of bias, due to serious imprecision12

We are uncertain whether roxadustat (any dose) improves QoL assessed by EQ-5D-5L VAS.

Difference: 1.42 higher (MD)

(CI 95% 1.21 lower–4.04 higher)

Fatigue measured by FACT—total score at 28 weeks

Measured by:

Scale: High better

Based on data from 783 participants in one study

Mean Mean

Very low

Due to very serious risk of bias, due to serious imprecision13

We are uncertain whether roxadustat (any dose) increases fatigue measured by FACT—total score at 28 weeks.

Difference: 2.41 higher (MD)

(CI 95% 1.68 lower–6.51 higher)

Incidence of MACE up to 6 weeks Based on data from 96 participants in one study No incidence of MACE was reported in either Roxadustat or ESA (epoetin alpha/darbepoetin alpha) group

Very low

Due to very serious risk of bias, due to serious indirectness, due to serious imprecision13

There were no participants who experienced MACE up to 6 weeks, so we were unable to determine whether roxadustat (any dose) made a difference.

Risk of Bias: serious. Inadequate/lack of blinding of participants and personnel, resulting in potential for performance bias, inadequate/lack of blinding of outcome assessors, resulting in potential for detection bias; Imprecision: serious. Wide confidence intervals, the 95% CI of the included study overlaps line of no effect (i.e., CI includes 1.0) rate; Publication bias: not serious. Mostly commercially funded studies. Risk of Bias: very serious. Inadequate/lack of blinding of participants and personnel, resulting in potential for performance bias, inadequate/lack of blinding of outcome assessors, resulting in potential for detection bias, selective outcome reporting; Imprecision: serious. Wide confidence intervals, wide confidence intervals, the 95% CI of the included study overlaps line of no effect (i.e., CI includes 1.0) rate; Publication bias: not serious. Mostly commercially funded studies. Risk of Bias: serious. Inadequate/lack of blinding of participants and personnel, resulting in potential for performance bias, inadequate/lack of blinding of outcome assessors, resulting in potential for detection bias; Inconsistency: serious. The magnitude of statistical heterogeneity was high with I^2 55 %; Imprecision: serious. Wide confidence intervals, the 95% CI of the included study overlaps line of no effect (i.e., CI includes 1.0) rate; Publication bias: not serious. Mostly commercially funded studies. Risk of Bias: very serious. Missing intention-to-treat analysis, inadequate/lack of blinding of participants and personnel, resulting in potential for performance bias, inadequate/lack of blinding of outcome assessors, resulting in potential for detection bias; Imprecision: very serious. Wide confidence intervals, the 95% CI of the included study overlaps line of no effect (i.e., CI includes 1.0) rate, low number of patients; Publication bias: not serious. Mostly commercially funded studies. Risk of Bias: very serious. Inadequate/lack of blinding of participants and personnel, resulting in potential for performance bias, inadequate/lack of blinding of outcome assessors, resulting in potential for detection bias, selective outcome reporting; Inconsistency: serious. The magnitude of statistical heterogeneity was high with I^2: 98%, the confidence interval of some of the studies do not overlap with those of most included studies/the point estimate of some of the included studies; Imprecision: serious. Wide confidence intervals, the 95% CI of the included study overlaps line of no effect (i.e., CI includes 1.0) rate; Publication bias: not serious. Mostly commercially funded studies. Risk of Bias: very serious. Inadequate/lack of blinding of participants and personnel, resulting in potential for performance bias, inadequate/lack of blinding of outcome assessors, resulting in potential for detection bias, selective outcome reporting; Imprecision: serious. Wide confidence intervals, the 95% CI of the included study overlaps line of no effect (i.e., CI includes 1.0) rate; Publication bias: not serious. Mostly commercially funded studies. Risk of Bias: serious. Inadequate/lack of blinding of participants and personnel, resulting in potential for performance bias, inadequate/lack of blinding of outcome assessors, resulting in potential for detection bias; Indirectness: serious. The included study was from countries other than South Asia and was downgraded for lack of directness by one level; Publication bias: not serious. Mostly commercially funded studies. Risk of Bias: very serious. Inadequate/lack of blinding of participants and personnel, resulting in potential for performance bias, inadequate/lack of blinding of outcome assessors, resulting in potential for detection bias, selective outcome reporting; Imprecision: serious. Wide confidence intervals, the 95% CI of the included study overlaps line of no effect (i.e., CI includes 1.0) rate, wide confidence intervals; Publication bias: not serious. Mostly commercially funded studies. Risk of Bias: very serious. Missing intention-to-treat analysis, inadequate/lack of blinding of participants and personnel, resulting in potential for performance bias, inadequate/lack of blinding of outcome assessors, resulting in potential for detection bias; Imprecision: serious. Due to less events (< 400) and inadequate optimal information size (OIS); Publication bias: not serious. Mostly commercially funded studies. Risk of Bias: serious. Inadequate/lack of blinding of participants and personnel, resulting in potential for performance bias, inadequate/lack of blinding of outcome assessors, resulting in potential for detection bias; Inconsistency: serious. The magnitude of statistical heterogeneity was high with I^2:56 %; Imprecision: serious. Wide confidence intervals, the 95% CI of the included study overlaps line of no effect (i.e., CI includes 1.0) rate; Publication bias: not serious. Mostly commercially funded studies. Risk of Bias: serious. Inadequate/lack of blinding of participants and personnel, resulting in potential for performance bias, inadequate/lack of blinding of outcome assessors, resulting in potential for detection bias; Publication bias: serious. Mostly commercially funded studies, asymmetrical funnel plot. Risk of Bias: very serious. Inadequate/lack of blinding of participants and personnel, resulting in potential for performance bias, inadequate/lack of blinding of outcome assessors, resulting in potential for detection bias, selective outcome reporting, missing intention-to-treat analysis; Imprecision: serious. Only data from one study, low number of patients, wide confidence intervals, the 95% CI of the included study overlaps line of no effect; Publication bias: not serious. Mostly commercially funded studies. Risk of Bias: very serious. Missing intention-to-treat analysis; inadequate/lack of blinding of participants and personnel, resulting in potential for performance bias, inadequate/lack of blinding of outcome assessors, resulting in potential for detection bias, selective outcome reporting; Imprecision: serious. Wide confidence intervals; low number of patients, only data from one study; Publication bias: not serious. Mostly commercially funded studies. Risk of Bias: very serious. Missing intention-to-treat analysis, inadequate/lack of blinding of participants and personnel, resulting in potential for performance bias, inadequate/lack of blinding of outcome assessors, resulting in potential for detection bias, incomplete data and/or large loss to follow-up; Indirectness: serious. The included study was not from South Asian country and was downgraded for lack of directness by one level; Imprecision: serious. Low number of patients, only data from one study; Publication bias: not serious. Mostly commercially funded studies. DD: Dialysis-Dependent, CKD: Chronic kidney disease, ESA: Eythropoiesis-stimulating agents, MACE: Major adverse cardiovascular events, HIF PHI: Hypoxia-inducible factor prolyl hydroxylase inhibitor, FACT: Functional assessment of cancer therapy (measure of fatigue), CI: Confidence interval, MD: Mean difference