Switching to Pegmolesatide in Maintenance Dialysis: Real-World Hemoglobin Response, Temporal Dynamics, and Modality-Specific Predictors

Abstract

Purpose

Pegmolesatide is a monthly subcutaneous erythropoietin mimetic peptide for renal anemia, but real-world data beyond randomized trials remain scarce. We evaluated its effectiveness, temporal response dynamics, and modality-specific response predictors in maintenance dialysis patients switched from prior anemia therapy.

Patients and Methods

In this single-center retrospective study, 192 maintenance dialysis patients (hemodialysis [HD], n = 90; peritoneal dialysis [PD], n = 102) receiving monthly subcutaneous pegmolesatide were analyzed; 52.6% had switched from erythropoiesis-stimulating agents and 47.4% from a hypoxia-inducible factor prolyl hydroxylase inhibitor, predominantly for suboptimal anemia control. The primary endpoint was hemoglobin (Hb) change at month 3. Secondary endpoints included iron dynamics, mixed-effects modeling of temporal patterns (1,423 observations), and predictors of poor response.

Results

Mean Hb increased from 96.4 ± 15.9 to 110.0 ± 19.2 g/L at month 3 (model-estimated Δ = +13.0 g/L; 95% CI, 9.8 to 16.3; P<0.001) and remained stable through month 12. Target achievement (Hb ≥ 110 g/L) rose from 19.8% to 54.5%. The Hb trajectory was nonlinear, with a rapid first-month rise followed by deceleration; increments were attenuated at higher baseline Hb (ceiling effect), and higher serum albumin was associated with faster Hb improvement. Iron deficiency declined from 46.3% at baseline to a model-estimated 20.4% at month 12. Higher baseline Hb predicted a smaller increment; dialysis vintage of 12–36 months independently predicted attenuated response (pooled OR 5.17–8.68; P<0.001). Predictors differed by modality (body mass index in HD; diabetes in PD). Thromboembolic events (2.6%) were exclusively vascular access-related; mortality was 2.1%.

Conclusion

Monthly pegmolesatide was associated with rapid and sustained Hb improvement and acceptable safety. Baseline Hb and dialysis vintage were the dominant response predictors, with distinct modality-specific profiles, supporting evaluation of individualized dosing strategies in randomized trials.

Introduction

Anemia is one of the most common and clinically important complications of chronic kidney disease (CKD), affecting over 90% of patients with end-stage renal disease (ESRD) and contributing to cardiovascular morbidity, diminished quality of life, and increased mortality.1,2 The pathophysiology of renal anemia is multifactorial, including relative erythropoietin deficiency, iron dysregulation mediated by hepcidin upregulation, shortened red blood cell lifespan, and chronic inflammation.1,3 Since the demonstration that recombinant human erythropoietin (rHuEPO) could correct the anemia of ESRD,4 erythropoiesis-stimulating agents (ESAs) have become the main treatment for anemia in dialysis patients. However, evidence from large clinical trials has raised concerns about aggressive ESA-driven hemoglobin (Hb) normalization, showing increased risk of thrombotic events, stroke, and mortality at higher targets.5–7 These findings led to more conservative Hb targets and prompted interest in alternative treatments for renal anemia.

Meanwhile, hypoxia-inducible factor (HIF) prolyl hydroxylase inhibitors (PHIs) have attracted attention because they stimulate endogenous erythropoietin production through physiological oxygen-sensing pathways.8–10 However, HIF activation has complex effects on iron metabolism, vascular function and glucose homeostasis, and its long-term safety is still being evaluated.11,12 Against this backdrop, erythropoietin mimetic peptides (EMPs) have emerged as a structurally distinct class of erythropoietic agents. Pegmolesatide, the most recently approved member of this class, comprises two identical EPO-mimetic peptide chains, originally identified through phage-display screening, conjugated to a branched polyethylene glycol (PEG) moiety, and acts as an agonist at the erythropoietin receptor (EPOR) through its EPO-mimetic peptide domains.13,14 The peptide shares no sequence homology with endogenous erythropoietin and consists entirely of natural amino acids, minimizing cross-reactivity with anti-erythropoietin antibodies; a β-alanine spacer at position 21 preserves receptor-binding affinity, and PEG conjugation extends the elimination half-life to approximately 70 hours (range 61.6–74.9 hours) in dialysis patients.15 Surface plasmon resonance studies show a 6.4–6.8-fold longer EPOR residence time than rHuEPO, allowing sustained activation of downstream JAK2/STAT5 and ERK1/2 MAPK signaling throughout dosing intervals.15 Phase 3 trials in non-dialysis-dependent CKD showed non-inferiority to epoetin alfa,16 and a dialysis study demonstrated comparable efficacy to epoetin alfa.17 Approved in China in June 2023 for anemia in both dialysis and non-dialysis CKD patients, pegmolesatide is currently the only marketed long-acting EPO-mimetic peptide; in Phase 2 studies it was administered subcutaneously once every 4 weeks at initial doses of 0.025–0.08 mg/kg, titrated every 4 weeks toward a target Hb of 10.0–12.0 g/dL.15

However, data on pegmolesatide efficacy in real-world practice remain limited. Randomized controlled trials (RCTs) enroll selected populations that may not reflect the real-world dialysis population in terms of comorbidity burden, dialysis vintage, iron status and treatment adherence. Long-term Hb response dynamics are also poorly understood in observational settings.12 No observational study has examined the time-interaction dynamics of pegmolesatide response in a dialysis cohort switched from prior ESA or HIF-PHI therapy. This study reports on 192 maintenance dialysis patients, all of whom had received prior ESA or HIF-PHI therapy before switching, predominantly for suboptimal anemia control. We examined temporal response patterns, iron metabolism dynamics, modality-specific predictors, and models for response stratification.

Materials and Methods

Study Design and Population

This retrospective real-world study was conducted at a single center comprising three campuses of the First Affiliated Hospital of Ningbo University, between January 2024 and December 2025. Data were extracted from the institutional electronic health record (EHR) system and dialysis database. The study protocol was approved by the ethics committee of the First Affiliated Hospital of Ningbo University (approval number: 2026–114) on March 27, 2026, prior to any data extraction and analysis; no data were accessed for research purposes before approval. The requirement for informed consent was waived given the retrospective nature of the analysis and the use of de-identified routinely collected clinical data. The study was conducted in accordance with the Declaration of Helsinki.

Inclusion Criteria

(i) age ≥18 years; (ii) maintenance hemodialysis (HD) or peritoneal dialysis (PD) for CKD stage 5; (iii) diagnosis of CKD-related anemia; (iv) receipt of at least one dose of pegmolesatide; (v) complete baseline clinical data and ≥1 month follow-up duration.

Exclusion Criteria

(i) a history of malignancy, active autoimmune disease, or primary hematologic disorder; (ii) severe infection, bleeding, or major surgery within 1 month prior to pegmolesatide initiation; (iii) non-renal anemia of any etiology; (iv) allogeneic blood transfusion within three months before enrollment; (v) concurrent use of other erythropoietic agents; (vi) combined HD and PD dialysis modality; (vii) missing critical data.

All enrolled 192 patients (HD 90, PD 102) had baseline Hb and ≥1 follow-up measurement, contributing 1,423 longitudinal observations.

Suboptimal anemia control was defined as Hb < 110 g/L despite ≥3 months of prior erythropoietic therapy at a stable dose, or physician-documented hyporesponsiveness to prior erythropoietic therapy.

Treatment Protocol

Pegmolesatide was administered as a subcutaneous injection at 4-week intervals, consistent with its extended pharmacokinetic profile enabled by polyethylene glycol conjugation.13,14 Taking drug accessibility and pharmacoeconomic considerations into account, the initial dose was generally 4 mg once every 4 weeks, with a maintenance dose range of 2–6 mg every 4 weeks. This range corresponds to the conversion dose bands recommended by the 2024 Chinese expert consensus on long-acting ESAs (2, 4, or 6 mg every 4 weeks, selected according to the prior weekly dose of short-acting ESA).18 The 4 mg starting dose was deliberately more conservative than the consensus-recommended band of 6 mg every 4 weeks for prior weekly short-acting ESA doses exceeding 6000 IU, permitting cautious individual up-titration in this switched population. Doses were reviewed at each 4-week visit and adjusted within this range, aiming to maintain Hb at ≥ 110 g/L while avoiding sustained concentrations above 130 g/L, as recommended by the Chinese renal anemia guideline.19 Concurrent iron supplementation (intravenous or oral) was prescribed at the physician’s discretion based on serial iron parameter evaluations.

Measurements and Assessments

Baseline variables included demographics (age, sex, body mass index [BMI]), dialysis modality (hemodialysis versus peritoneal dialysis), dialysis vintage (months), comorbidities (diabetes mellitus, cardiovascular disease [CVD]), prior ESA or HIF-PHI use, and vascular access type for hemodialysis patients (arteriovenous fistula or tunneled catheter). Blood pressure (systolic and diastolic; mmHg) was measured at each visit using standardized techniques.

Laboratory assessments were performed at 13 scheduled timepoints. Hematological parameters included Hb (g/L), measured by automated hematology analyzer. Iron metabolism parameters included serum iron (μmol/L), total iron-binding capacity (TIBC; μmol/L), transferrin saturation (TSAT; %), and ferritin (ng/mL). Inflammatory and nutritional markers included high-sensitivity C-reactive protein (hs-CRP; mg/L, with >5 mg/L defining clinically significant inflammation), serum albumin (Alb; g/L), serum creatinine (Scr; μmol/L), and electrolytes (potassium, mmol/L). Lipid profiles included total cholesterol (TC; mmol/L), triglycerides (TG; mmol/L), low-density lipoprotein cholesterol (LDL-C; mmol/L), and high-density lipoprotein cholesterol (HDL-C; mmol/L). Mineral and bone metabolism parameters included intact parathyroid hormone (PTH; pg/mL). Dialysis adequacy was assessed using single-pool Kt/V for hemodialysis patients and weekly Kt/V and creatinine clearance (CCr; L/week/1.73 m²) for peritoneal dialysis patients.

Iron deficiency was categorized into three types: absolute iron deficiency (TSAT < 20% and ferritin < 100 ng/mL), functional iron deficiency (TSAT < 20% with ferritin ≥ 100 ng/mL), and sufficient iron status (TSAT ≥ 20%).20,21 A broader criterion (TSAT < 20% or ferritin < 100 ng/mL) was adopted for the primary analysis.3

Statistical Analysis

The primary endpoint was Hb change from baseline to month 3. Secondary endpoints included: (i) Hb at months 1, 3, 6, 9, and 12; (ii) target achievement rate (Hb ≥ 110 g/L), including time to first attainment and proportion of subsequent observations maintaining the target after initial achievement; (iii) iron parameter dynamics; (iv) temporal response patterns; and (v) predictors of poor response. Continuous variables were summarized as mean ± SD or median [IQR]; categorical variables as frequencies and percentages. Between-group comparisons used t-tests or Mann–Whitney U-tests (continuous) and chi-squared or Fisher’s exact tests (categorical). Paired comparisons over time used Wilcoxon signed-rank tests. Between-modality Hb comparisons used Mann–Whitney U-tests; cumulative distribution functions were compared using the Kolmogorov–Smirnov test. Longitudinal analysis of iron deficiency prevalence used a generalized estimating equation (GEE) logistic model with an exchangeable within-patient correlation structure and robust standard errors, including all six timepoints (baseline, months 1, 3, 6, 9, and 12) as categorical indicators with baseline as the reference. As a sensitivity analysis, pairwise baseline-versus-follow-up comparisons were performed using McNemar tests (with continuity correction, or exact tests when the number of discordant pairs was < 25) restricted to patients with paired measurements.

Hb trajectories were analyzed using linear mixed-effects models (LMM) with restricted maximum likelihood (REML) estimation, random intercept at the patient level, and quadratic time (month + month²). The model incorporated 1,423 observations from 192 patients. Primary contrasts used Bonferroni correction (k = 5; α = 0.010). Four temporal interactions were assessed: (i) baseline Hb-time interaction (ceiling effects); (ii) dose-time interaction (attenuation of the dose association over time); (iii) sex-time interaction; and (iv) time-varying albumin and hs-CRP (nutritional and inflammatory associations). To characterize inter-individual heterogeneity in response trajectories, patients were classified into three responder groups (rapid, moderate, slow) based on individual slope estimates from patient-specific linear regressions requiring ≥2 post-baseline Hb measurements, with cutpoints at the 25th and 75th percentiles of the slope distribution. Intra-individual Hb variability was quantified as the within-patient coefficient of variation across follow-up measurements.

Thirteen pre-specified subgroup analyses examined treatment effect heterogeneity by baseline Hb, age, sex, BMI, diabetes, hs-CRP, prior ESA, dialysis modality, albumin, PTH, dialysis vintage, iron deficiency, and CVD. After Benjamini-Hochberg false discovery rate (FDR) correction, baseline Hb and dialysis vintage remained significant at q < 0.05.

Predictive modeling of poor response used univariable screening at P<0.20 followed by multivariable backward elimination guided by the Akaike information criterion (AIC). Restricted cubic spline (RCS) analysis revealed a significant non-linear relationship with month-3 Hb change (likelihood ratio test for non-linearity: χ² = 18.43, P<0.0001), with the RCS curve reaching its minimum at 24.2 months. Based on this empirical U-shaped pattern and clinical considerations, dialysis vintage was categorized into three groups (<12, 12–36, and ≥36 months). This categorization was supported by AIC comparison (categorical model AIC 1376.6 versus continuous model AIC 1388.7, ΔAIC = −12.1) and was consistent across both HD and PD subgroups. RCS analysis of other continuous variables (baseline Hb, age, BMI, ferritin, TSAT, serum iron, CRP, PTH, albumin) revealed no significant non-linear relationships with month-3 Hb change (all P>0.05 for non-linearity). Six models were constructed: two pooled logistic (Model A: Hb₃ < 100 g/L; Model B: ΔHb₃ < 5 g/L), two HD-specific, and two PD-specific (for each modality, a linear model for ΔHb₃ and a logistic model for ΔHb₃ < 5 g/L). Discrimination was assessed by AUC with 95% CIs (DeLong test). Events per variable (EPV) were 9.0 (Model A) and 7.4 (Model B); Firth penalized regression served as sensitivity analysis. Complete-case analysis was used for outcome assessments. Missing data for month 3 hemoglobin: HD, 21.1% (90 to 71); PD, 16.7% (102 to 85); pooled cohort, 18.8% (192 to 156). Missing baseline covariates were median-imputed (iron parameters 1–2%; albumin and PTH 0.5–1.4%). Sensitivity analyses excluding imputed cases were materially unchanged. All final-model variance inflation factor (VIF) values were below 2.0 (range 1.15–1.57), indicating no multicollinearity concern. As a temporal sensitivity analysis, the identical variable selection procedure was repeated using month-1 hemoglobin change (ΔHb₁) as a continuous outcome to assess whether month-3 predictors also governed early response.

All analyses were performed using R version 4.3.3. Two-sided P<0.05 was considered significant unless otherwise specified.

Results

Baseline Characteristics

A total of 933 maintenance dialysis patients were screened during the study period, of whom 192 satisfied all predefined inclusion and exclusion criteria and were included in the final analysis (Figure 1).

Figure 1.

Flow diagram of patient selection and grouping.

All 192 patients had received prior anemia pharmacotherapy before pegmolesatide initiation: 101 (52.6%) were receiving erythropoiesis-stimulating agents (recombinant human erythropoietin; mean weekly epoetin-equivalent dose 12,574.3 ± 3,647.9 IU) and 91 (47.4%) were receiving the hypoxia-inducible factor prolyl hydroxylase inhibitor roxadustat (median dose 450 [IQR 300–450] mg/week). The median duration of prior anemia therapy was 8.5 (IQR 5.6–12.8) months. Reasons for switching to pegmolesatide were not mutually exclusive and included inadequate anemia control in 154 patients (80.2%), inconvenient prior dosing regimens or physician/patient preference in 45 (23.4%), and adverse effects or other reasons in 7 (3.6%).

The baseline characteristics are presented in Table 1. The mean age was 58.7 ± 13.1 years, with HD patients being significantly older than PD patients (62.9 ± 12.5 versus 55.0 ± 12.4 years; P<0.001). Male sex predominated (60.9% overall) without significant between-modality differences (HD 61.1% versus PD 60.8%; P=0.964). BMI was comparable between groups (23.0 ± 3.6 kg/m² overall; P=0.254).

Table 1.

Baseline Demographic and Clinical Characteristics of the Study Cohort, Stratified by Dialysis Modality

Variable	Overall (n=192)	HD (n=90)	PD (n=102)	P-value
Age, years	58.7 ± 13.1	62.9 ± 12.5	55.0 ± 12.4	<0.001
Male sex, n (%)	117 (60.9)	55 (61.1)	62 (60.8)	0.964
BMI, kg/m2	23.0 ± 3.6	22.7 ± 3.3	23.3 ± 3.8	0.254
Dialysis vintage, months	15.5 [5.2, 41.6]	20.7 [7.3, 36.8]	11.2 [2.8, 45.2]	0.094
Diabetes, n (%)	81 (42.2)	48 (53.3)	33 (32.4)	0.003
CVD, n (%)	54 (28.1)	38 (42.2)	16 (15.7)	<0.001
Prior ESA, n (%)	101 (52.6)	65 (72.2)	36 (35.3)	<0.001
Baseline Hb, g/L	96.4 ± 15.9	92.5 ± 13.5	99.9 ± 17.1	<0.001
TSAT, %	31.6 ± 17.0	30.9 ± 16.8	32.2 ± 17.2	0.796
Ferritin, ng/mL	220.8 ± 218.8	258.3 ± 248.7	187.6 ± 175.6	0.050
Serum iron, μmol/L	13.3 ± 6.4	12.8 ± 6.6	13.7 ± 6.2	0.268
PTH, pg/mL	236.3 ± 175.7	196.6 ± 107.8	272.7 ± 214.6	0.103
hs-CRP >5 mg/L, n (%)	55 (28.6)	33 (36.7)	22 (21.6)	0.030
Iron deficiency, n (%)	87 (46.3)	40 (45.5)	47 (47.0)	0.948
Avg. dose (first 3 mo), mg	3.93 ± 0.91	4.16 ± 1.00	3.73 ± 0.77	<0.001
Target Hb ≥ 110 g/L, n (%)	38 (19.8)	12 (13.3)	26 (25.5)	0.038

Notes: Data are presented as mean ± SD, median [interquartile range], or n (%) as appropriate. Average dose represents the mean monthly subcutaneous pegmolesatide dose (mg) administered during the first 3 months of therapy. Iron deficiency prevalence is based on patients with available baseline TSAT and ferritin measurements (overall n = 188; HD n = 88; PD n = 100).

Abbreviations: BMI, body mass index; CVD, cardiovascular disease; ESA, erythropoiesis-stimulating agent; Hb, hemoglobin; TSAT, transferrin saturation; PTH, parathyroid hormone; hs-CRP, high-sensitivity C-reactive protein.

HD patients exhibited a substantially greater burden of comorbid disease: diabetes was more prevalent (53.3% versus 32.4%; P=0.003), as was CVD (42.2% versus 15.7%; P<0.001). Prior ESA exposure was also significantly higher in the HD group (72.2% versus 35.3%; P<0.001). Correspondingly, baseline Hb was lower in HD patients (92.5 ± 13.5 versus 99.9 ± 17.1 g/L; P<0.001), and the proportion achieving the ≥ 110 g/L target was lower (13.3% versus 25.5%; P=0.038). Elevated inflammation (hs-CRP >5 mg/L) was more common in HD patients (36.7% versus 21.6%; P=0.030). Iron deficiency prevalence was high overall (46.3%) without significant between-modality differences (P=0.948). The average dose during the first 3 months was significantly higher in HD patients (4.16 ± 1.00 versus 3.73 ± 0.77 mg; P<0.001), consistent with their lower baseline Hb and greater ESA utilization.

Hemoglobin Response Over 12 Months

The primary endpoint (change in Hb from baseline to month 3) was achieved with clinically and statistically significant improvements (Table 2; Figure 2a–d). Mean Hb levels increased from 96.4 ± 15.9 g/L at baseline to 110.0 ± 19.2 g/L at month 3, with a model-estimated mean increase of +13.0 g/L (95% CI, 9.8 to 16.3; Bonferroni-corrected P<0.001) (Figure 2a). An early response was already apparent at month 1 (+9.6 g/L; 95% CI, 7.6 to 11.7; P<0.001).

Table 2.

Longitudinal Hemoglobin Response to Pegmolesatide Over 12 Months

Timepoint	N	Hb, g/L (mean ± SD)	ΔHb vs baseline (95% CI)	Bonferroni P	Target Hb ≥ 110 g/L, %
Baseline	192	96.4 ± 15.9	–	–	19.8
Month 1	192	106.0 ± 17.8	+9.6 (7.6, 11.7)	<0.001	40.6
Month 3	156	110.0 ± 19.2	+13.0 (9.8, 16.3)	<0.001	54.5
Month 6	117	107.7 ± 15.2	+10.4 (7.4, 13.4)	<0.001	54.7
Month 9	70	109.4 ± 14.3	+13.1 (8.8, 17.4)	<0.001	48.6
Month 12	31	110.6 ± 18.2	+15.6 (7.7, 23.6)	0.002	51.6

Notes: ΔHb represents marginal mean differences from baseline estimated by linear mixed-effects models, with 95% confidence intervals. P-values are Bonferroni-corrected for multiple comparisons.

Abbreviation: Hb, hemoglobin.

Figure 2.

Hemoglobin response to pegmolesatide over 12 months. (a) Overall mean Hb trajectory with SEM; Wilcoxon signed-rank tests vs baseline: **P<0.01, ***P<0.001. Dashed lines: Hb target ≥ 110 g/L (red) and initiation threshold 100 g/L (orange). (b) Hb trajectories by dialysis modality; asterisks indicate between-group comparisons (HD vs PD) at the corresponding time points by Mann–Whitney U-test: *P<0.05, **P<0.01. (c) Target achievement rate (Hb ≥ 110 g/L); dashed line indicates the 50% achievement level. (d) Cumulative distribution function of Hb at month 3 by dialysis modality; two-sample Kolmogorov–Smirnov test: D=0.215, *P<0.05.

Abbreviations: HD, hemodialysis; PD, peritoneal dialysis; SEM, standard error of the mean; BL, baseline; M, month.

The target achievement rate increased substantially from baseline, with 40.6% of patients reaching Hb ≥ 110 g/L at month 1, rising to 54.5% at month 3 (Figure 2c). The increment at month 6 (+10.4 g/L; 95% CI, 7.4 to 13.4; P<0.001) was modestly attenuated relative to the month 3 peak before returning to a comparable magnitude at month 9 (+13.1 g/L; 95% CI, 8.8 to 17.4; P<0.001), suggesting stabilization after the initial rapid response phase. At month 12, the available subset (n=31) demonstrated a mean Hb of 110.6 ± 18.2 g/L (Δ = +15.6 g/L; 95% CI, 7.7 to 23.6; P=0.002), with 51.6% achieving the ≥ 110 g/L target. Of 192 patients, 138 (71.9%) achieved Hb ≥ 110 g/L at least once during follow-up (median time to first attainment: 1 month; IQR 0–3). Target maintenance after first attainment averaged 76.9% of subsequent observations (median 85.7%).

Hemodialysis patients started from a lower baseline (92.5 ± 13.5 versus 99.9 ± 17.1 g/L) but exhibited a comparable absolute increment at month 3, narrowing the between-modality gap from 7.4 g/L at baseline to 4.1 g/L at month 3 (Figure 2b and d). The pattern was consistent across timepoints: at month 6, HD patients maintained a mean Hb of 105.8 g/L while PD patients reached 109.1 g/L, preserving a modest but stable modality differential despite equivalent dosing intervals.

Of the 192 patients, 65 were enrolled on or before January 2025 and were therefore eligible for 12-month follow-up at the analysis cutoff; 31 (47.7%) completed the month-12 assessment. Reasons for non-completion (n = 34; HD 21, PD 13) were transfer to another facility or loss to follow-up (n = 12), kidney transplantation (n = 2), patient refusal to continue (n = 5), inadequate response with switch to alternative therapy (n = 5), death (n = 1), and other reasons (n = 9). Non-completers exhibited early hemoglobin responses comparable to the overall cohort (baseline, 97.1 ± 14.4 g/L; month 1, 106.9 ± 19.1 g/L; month 3, 113.9 ± 24.4 g/L; P<0.05 versus baseline at both timepoints), indicating that attrition was not driven by treatment failure.

Iron Metabolism Dynamics

Iron metabolism parameters varied across the 12-month follow-up (Figure 3a–f). Transferrin saturation increased from a baseline of 31.6 ± 17.0% to 36.8 ± 15.3% at month 3 (Δ = +5.2; P<0.001) and further to 41.8 ± 16.1% at month 12 (Δ = +10.2; P=0.059) (Figure 3a), suggesting a trend toward improved functional iron availability. Serum iron showed a directional increase from 13.3 ± 6.4 to 15.1 ± 6.1 μmol/L at month 3 (Δ = +1.8; P=0.018). Total iron-binding capacity decreased significantly at all evaluated timepoints (month 3: Δ = −3.5; P<0.001; month 12: Δ = −9.6; P=0.004). Ferritin was 220.8 ± 218.8 ng/mL at baseline, remained stable at month 3 (231.6 ± 211.9 ng/mL; P=0.963), and increased by month 12 (361.1 ± 290.7 ng/mL; P<0.001).

Figure 3.

Iron metabolism dynamics under pegmolesatide. (a) TSAT trajectory by baseline iron status. (b) Iron deficiency prevalence over time (absolute: TSAT < 20% and ferritin < 100 ng/mL; functional: TSAT < 20% and ferritin ≥ 100 ng/mL; sufficient: TSAT ≥ 20%). (c and d) Paired boxplots of iron parameters at baseline versus month 3 for HD (c) and PD (d). (e) Baseline ferritin versus Hb change at month 3 (Spearman rho=+0.177, P=0.028). (f) Ferritin trajectory by IV iron status.

Abbreviations: TSAT, transferrin saturation; TIBC, total iron-binding capacity; HD, hemodialysis; PD, peritoneal dialysis; IV, intravenous.

The prevalence of iron deficiency declined over follow-up (Figure 3b). In the GEE model (679 observations from 192 patients), the odds of iron deficiency were significantly lower than baseline from month 3 onward (month 3: OR 0.63, 95% CI 0.44–0.90, P=0.012; month 6: OR 0.45, 95% CI 0.28–0.70, P<0.001; month 9: OR 0.43, 95% CI 0.25–0.74, P=0.002; month 12: OR 0.30, 95% CI 0.11–0.82, P=0.019), with model-estimated prevalences of 35.0% at month 3 and 20.4% at month 12 versus 46.3% at baseline. Paired McNemar sensitivity analyses were consistent in direction and magnitude (month 3: 43.1% to 34.0%, n = 153 pairs, P=0.066; month 12: 57.7% to 19.2%, n = 26 pairs, exact P=0.013). Between baseline and month 3, absolute deficiency decreased from 20.7% to 7.1% (P<0.001), functional deficiency decreased from 6.9% to 5.2% (P=0.360), and sufficiency (TSAT ≥20%) increased from 72.3% to 87.7% (P=0.002). These changes were consistent in both modalities (Figure 3c and d). Baseline ferritin correlated weakly with month-3 Hb change (rho = +0.177; P=0.028) (Figure 3e); TSAT quartile stratification was not significant (P=0.679).

During the first 3 months, 47 patients (24.5%) received intravenous iron supplementation, predominantly in HD patients (HD 41.1% versus PD 9.8%; P<0.001), while oral iron was received by 81.4% of PD patients. Baseline ferritin did not differ between patients who received intravenous iron during the first 3 months and those who did not (202.1 ± 219.5 versus 225.7 ± 219.1 ng/mL; P=0.551); at month 3, ferritin was higher in intravenous iron recipients (367.1 ± 299.4 versus 195.2 ± 165.0 ng/mL; P<0.001) (Figure 3f).

Temporal Response Dynamics: Interaction Patterns and Nonlinear Trajectories

Longitudinal mixed-effects modeling (1,423 observations from 192 patients across 13 timepoints) revealed a nonlinear hemoglobin trajectory and four distinct temporal interaction patterns that characterize the dynamics of pegmolesatide response.

Nonlinear Time Trend

The overall Hb trajectory followed a quadratic pattern (Time² β = −0.232; P<0.001), with an initial rapid rise (induction phase: +4.3 g/L/month), followed by deceleration (a slight decline from 110.0 g/L at month 3 to 107.7 g/L at month 6) and stabilization around 108–110 g/L. The M0-M1 interval showed the largest gain (+9.6 g/L).

Pattern 1

Hemoglobin ceiling effect. The baseline Hb-time interaction was significantly negative (β = −0.038 per month per 1 g/L baseline Hb; P<0.001), indicating that patients with higher baseline Hb experienced progressively attenuated incremental responses. The ceiling effect intensified over time: the attenuation in Hb gain per 1 g/L higher baseline Hb widened from 0.038 g/L at month 1 to 0.23 g/L by month 6 (equivalent to approximately 4.6 g/L by month 6 for a 20 g/L baseline difference).

Pattern 2

Attenuation of the negative dose association. The pegmolesatide association with Hb was consistently negative, consistent with confounding by indication (higher doses prescribed to poorer responders), and this association weakened progressively over time (time-dose interaction, P=0.002). The negative dose coefficient fell from −3.76 at month 1 to −1.96 at month 6, corresponding to a 48% attenuation.

Pattern 3

Emerging male advantage. The sex-time interaction was significantly positive (β = +0.758 per month; P<0.001), showing a progressive male advantage over the treatment course. The gap attributable to this interaction (relative to baseline) increased from 0.76 g/L at month 1 to 2.27 g/L at month 3 and 4.55 g/L at month 6.

Pattern 4

Nutritional status and inflammation. Time-varying serum albumin was positively associated with Hb (β = +1.22 per g/L; P<0.0001), with a positive albumin-time interaction (β = +0.053 per month; P=0.035), indicating that better nutritional status was associated with both higher absolute Hb levels and faster Hb improvement over time. Conversely, time-varying log-transformed hs-CRP was negatively associated with Hb (β = −2.37; P<0.0001). The hs-CRP-time interaction was not significant (P=0.350).

Responder Trajectory Classification

Based on individual slope estimates from patient-specific linear regressions (requiring ≥2 post-baseline Hb measurements), three trajectory types were identified among 165 patients with sufficient longitudinal data: rapid responders (individual slope >75th percentile; n = 41, 24.8%), moderate responders (25th-75th percentile; n = 83, 50.3%), and slow responders (<25th percentile; n = 41, 24.8%). Rapid responders showed steep rises with plateauing (mean slope +4.6 g/L/month; Time² β = −0.343; P<0.001), whereas slow responders had near-flat trajectories (mean slope −0.9 g/L/month) and higher baseline Hb (102.4 versus 99.5 g/L; Time² β = +0.109; P=0.017). The responder × time interaction confirmed distinct longitudinal dynamics (β = +5.43; P<0.001). The within-patient coefficient of variation in Hb averaged 10.6% (median 9.8%; range 2.7–31.6%); baseline Hb was significantly associated with intra-individual variability (Spearman rho = −0.405; P<0.001), with patients at lower baseline Hb exhibiting greater within-subject fluctuations during follow-up. Serum albumin increased modestly from baseline to month 3 (Δ = +1.22 g/L; P<0.001); changes in creatinine, lipid profiles, electrolytes, and PTH were small and not clinically meaningful.

Subgroup Analysis and Treatment Effect Heterogeneity

Subgroup analysis revealed five significant differences in month-3 Hb (Figure 4a): baseline Hb <95 vs ≥95 g/L (−7.9 g/L; P=0.013), BMI <25 vs ≥25 (+7.1 g/L; P=0.047), HD vs PD (−6.9 g/L; P=0.023), hs-CRP >5 vs ≤5 mg/L (−10.3 g/L; P=0.004), and dialysis vintage <12 vs ≥12 months (+12.9 g/L; P<0.001). No significant differences were observed for age, sex, diabetes, CVD, prior ESA, iron status, albumin, or PTH (all P≥0.085).

Figure 4.

Forest plot of subgroup analyses of hemoglobin response. (a) Cross-sectional analysis at month 3: mean differences in absolute Hb (95% CI) for 13 baseline characteristics. (b) Longitudinal analysis: mean differences in ΔHb (95% CI) for 8 characteristics. Red indicates higher Hb in the first-named group (P<0.05); blue indicates lower Hb (P<0.05); gray indicates P≥0.05. FDR q<0.05 (Benjamini-Hochberg).

Abbreviations: hs-CRP, high-sensitivity C-reactive protein; HD, hemodialysis; PD, peritoneal dialysis.

For longitudinal ΔHb (Figure 4b), two subgroups differed: baseline Hb <95 g/L showed a larger increase (+17.5 g/L; P<0.001) and dialysis vintage <12 months showed a larger increase (+11.7 g/L; P<0.001). No other subgroup showed significant differences in ΔHb (all P≥0.063).

Predictors of Hemoglobin Response at Month 3

To identify factors influencing the Hb response to pegmolesatide at month 3, six multivariable models were established for the overall cohort and each dialysis modality (Figure 5a–f).

Figure 5.

Forest plot of six multivariable regression models for hemoglobin response at month 3. (a) Logistic Model A for Hb₃ < 100 g/L in the overall cohort (n=156, AUC=0.848); (b) logistic Model B for ΔHb₃ < 5 g/L in the overall cohort (n=156, AUC=0.852); (c) linear model for ΔHb₃ in hemodialysis (HD) patients (n=71, adjusted R²=0.313); (d) linear model for ΔHb₃ in peritoneal dialysis (PD) patients (n=85, adjusted R²=0.318); (e) logistic model for ΔHb₃ < 5 g/L in HD patients (n=71, AUC=0.900); (f) logistic model for ΔHb₃ < 5 g/L in PD patients (n=85, AUC=0.859). Odds ratios (log scale) for logistic models (a, b, e and f) and regression coefficients for linear models (c and d) are displayed with 95% confidence intervals. Red indicates risk association (OR>1 or β>0); blue indicates protective association (OR<1 or β<0). Dark colors: P<0.05; light colors: P≥0.05 (trend). Vintage reference: <12 months.

Pooled Predictive Models (Model A and Model B)

Model A (n = 156, AUC = 0.848) predicted Hb₃ < 100 g/L (Figure 5a). Baseline Hb was the strongest protective factor (OR 0.94, 95% CI 0.91–0.97, P<0.001). BMI (OR 1.20, P=0.006) and cumulative pegmolesatide dose (OR 1.94, P=0.019) were significant risk factors. Dialysis vintage 12–36 months showed the strongest risk association (OR 8.68, 95% CI 3.51–21.47, P<0.001). TSAT (P=0.221), PTH (P=0.082), and hs-CRP (P=0.811) were not significant.

Model B (n = 156, AUC = 0.852) predicted poor Hb increment (ΔHb₃ < 5 g/L) (Figure 5b). Baseline Hb was associated with a higher probability of poor incremental response (OR 1.08, 95% CI 1.04–1.12, P<0.001). Conversely, older age exerted a protective effect (OR 0.97, P=0.042). BMI (OR 1.17, P=0.016) and dialysis vintage 12–36 months (OR 5.17, P<0.001) remained significant, consistent with Model A. Higher TSAT also conferred a protective effect (OR 0.96, P=0.043), whereas hs-CRP (P=0.455) and ferritin (P=0.445) were not significant. Sensitivity analysis using Firth’s penalized regression generated consistent estimates.

Dialysis Modality-Specific Predictors

In HD subgroup, the linear regression model (n = 71, adjusted R² = 0.313) identified baseline Hb (β = −0.63, P<0.001) and dialysis vintage 12–36 months (β = −12.37, P=0.004) as significant predictors (Figure 5c). The logistic model for poor Hb increment (n = 71, AUC = 0.900) confirmed baseline Hb (OR 1.11, P=0.004), BMI (OR 1.38, P=0.029) and dialysis vintage 12–36 months (OR 8.06, P=0.009) as independent risk factors (Figure 5e). Additionally, TSAT exhibited a borderline protective trend toward better treatment outcomes (OR = 0.95, P=0.054).

In PD subgroup, the linear regression model (n = 85, adjusted R² = 0.318) identified CVD (β = 12.86, P=0.018; not observed in HD), dialysis vintage 12–36 months (β = −10.59, P=0.040), and baseline Hb (β = −0.56, P<0.001) as significant predictors (Figure 5d). The logistic model for poor Hb increment (n = 85, AUC = 0.859) revealed diabetes (OR 3.75, P=0.045) and baseline Hb (OR 1.07, P=0.005) as independent predictors (Figure 5f). Dialysis vintage failed to enter the final PD logistic model.

Sensitivity Analysis: Early Response Predictors at Month 1

To assess whether month-3 response predictors also govern early effects, the models were repeated using month-1 hemoglobin change (ΔHb1) as the outcome. Core findings were concordant: baseline Hb remained the strongest predictor (β = −0.286, P<0.001), dialysis vintage of 12–36 months was associated with attenuated early response (β = −7.19, P=0.001), and the negative dose coefficient (β = −253.3 per mg/kg, P<0.001) again reflected confounding by indication. Two modality-specific early predictors emerged: intravenous iron administration during the first month in hemodialysis patients (β = +7.54, 95% CI 2.74 to 12.34, P=0.002; n = 86) and peritoneal Kt/V in peritoneal dialysis patients (β = +14.33, 95% CI 5.73 to 22.94, P=0.001; n = 100), the latter suggesting that dialysis adequacy facilitates early erythropoiesis. Neither association remained significant at month 3.

Safety Outcomes

Safety outcomes are summarized in Table 3. Total follow-up comprised 1,294 person-months (107.8 person-years) across 192 patients, with a median follow-up of 7 months (IQR 3–12 months). Event rates per 100 person-years were: mortality 3.7/100 person-years (4 events), cardiovascular events 8.3/100 person-years (9 events), thromboembolic events 4.6/100 person-years (5 events), and hospitalization 77.0/100 person-years (83 patients with at least one hospitalization). Hospitalization occurred in 83 of 192 patients (43.2%) during follow-up. Non-access-related hospitalizations accounted for 34.4% of the total cohort (66 patients). There was no significant difference in hospitalization rates between HD and PD patients (47.8% versus 39.2%; P=0.252).

Table 3.

Safety Outcomes and Management Interventions During 12-month Follow-Up

Outcome	Overall (n=192)	HD (n=90)	PD (n=102)	P-value
Hospitalization, n (%)	83 (43.2)	43 (47.8)	40 (39.2)	0.252
Non-access hospitalization, n (%)	66 (34.4)	–	–	–
Thromboembolic events, n (%)	5 (2.6)	5 (5.6)	0 (0)	0.023
Cardiovascular events, n (%)	9 (4.7)	5 (5.6)	4 (3.9)	0.487
Death, n (%)	4 (2.1)	3 (3.3)	1 (1.0)	0.319

Five patients (2.6%) experienced thromboembolic events, all of which were vascular access-related (arteriovenous fistula thrombosis requiring interventional therapy). No systemic thromboembolic events (stroke, myocardial infarction, or deep vein thrombosis) were observed. All 5 events occurred in HD patients, representing 5.6% of the HD cohort. No thromboembolic events were recorded in PD patients. Nine patients (4.7%) experienced cardiovascular events, including 4 deaths (2.1%): 2 intracerebral hemorrhages, 1 acute cerebral infarction, and 1 acute myocardial infarction. Among the 5 non-fatal cardiovascular events, 3 were acute coronary syndromes and 2 were cerebrovascular accidents. The cardiovascular event rate was comparable between HD (5.6%) and PD (3.9%) patients (P=0.487). No significant alterations in systolic (−2.9 ± 14.5 mmHg; P=0.137) or diastolic blood pressure (+0.3 ± 9.2 mmHg; P=0.763) were observed at month 3.

Discussion

This 12-month retrospective study of 192 maintenance dialysis patients identified five principal findings. First, pegmolesatide produced a rapid and sustained Hb increase, with target achievement rates increasing from 19.8% to 54.5% at month 3. Second, functional iron parameters also improved. Intravenous iron was associated with greater early (month-1) Hb gain in HD patients but showed no independent association with month-3 response. Third, longitudinal modeling revealed four temporal interaction patterns, a hemoglobin ceiling effect, attenuation of the negative dose association, an emerging male advantage, and nutritional-inflammation associations, together stratifying patients into three distinct responder trajectories. Fourth, baseline Hb and dialysis vintage of 12–36 months were the strongest predictors of response. Baseline Hb showed a ceiling effect across all six models, while dialysis vintage 12–36 months showed a U-shaped relationship. Predictors differed between hemodialysis and peritoneal dialysis. Fifth, safety outcomes were acceptable: thromboembolic events were limited to vascular access-related complications, no blood pressure elevation was observed, and cardiovascular and mortality rates were within expected ranges.

In the Phase 3 dialysis trial by Zhang et al,17 participants were stable maintenance patients (mean baseline Hb 11.1 g/dL, within the 10.0–12.0 g/dL target) who achieved equivalent 24-week efficacy to epoetin alfa (+0.07 versus −0.22 g/dL). In contrast, the non-dialysis trial enrolled patients below target (Hb <100 g/L) and reported a larger treatment effect (+19.2 versus +15.4 g/L versus epoetin alfa),16 showing that incremental Hb gain is inversely related to baseline level. Our switched cohort had a mean baseline Hb of 96.4 g/L, with a 13.0 g/L increase at month 3, corresponding to an intermediate absolute treatment response. The 54.5% target attainment rate (Hb ≥ 110 g/L) in our study, while lower than the 67.8% target-range maintenance rate reported in the dialysis trial,17 reflects fundamentally different outcome definitions (single-threshold versus range-based) and patient selection (refractory-switch versus stable-maintenance), so direct comparison is not appropriate.

Comparable increments have been reported with other erythropoietic agents. In the phase 3 dialysis trial of roxadustat, mean Hb increased by 0.7±1.1 g/dL from a baseline of approximately 10.4 g/dL,8 and in ASCEND-D, daprodustat was non-inferior to ESAs in 2,964 dialysis patients with Hb maintained within the target range.9 Within the same class of long-acting agents, once-monthly C.E.R.A. (methoxy polyethylene glycol-epoetin beta) likewise maintained stable Hb in dialysis patients converted from shorter-acting ESAs: in PROTOS, monthly subcutaneous C.E.R.A. was non-inferior to epoetin administered one to three times weekly,22 and the correction-phase AMICUS trial, the only study in this class to enroll both HD and PD patients, reported similar Hb increments with intravenous C.E.R.A. every 2 weeks versus epoetin three times weekly (2.70 vs 2.56 g/dL).23 Its larger correction-phase increments reflect ESA-naïve enrollment and protocolized up-titration toward an 11–13 g/dL target, in contrast to the conservative real-world titration and 130 g/L ceiling in our cohort. In a pooled analysis of 13 phase 3 trials (2,060 patients switched to once-monthly C.E.R.A), 78% of patients maintained stable Hb with only moderate fluctuations.24 Notably, these maintenance-phase studies enrolled patients already stable at target (baseline Hb within 10.0–13.0 g/dL target ranges) with adequate iron status and were designed to maintain, rather than increase, Hb. The larger increment in our cohort is consistent with the inverse relationship between baseline Hb and achievable gain, and with the lower baseline Hb of this switched population, rather than indicating differential drug potency; differences in iron management protocols, endpoint definitions, and patient selection preclude direct numerical comparison.

A more useful comparison is the iron metabolic profile. The phase 3 trials mandated iron repletion at entry (ferritin ≥ 100 ng/mL, TSAT ≥ 20%) and allowed IV iron supplementation,16,17 conditions that do not reflect real-world practice where 46.3% of patients presented with iron deficiency. Despite the suboptimal iron status, pegmolesatide produced a distinct iron metabolic profile. TSAT rose and TIBC fell significantly, whereas ferritin remained stable at month 3 before increasing by month 12. This pattern of enhanced functional iron delivery with no early depletion of iron stores differs from HIF-PHIs (eg, roxadustat) that mobilize iron by suppressing hepcidin and reducing ferritin,8,12 as well as conventional ESAs that frequently deplete iron reserves during sustained erythropoiesis. Such metabolic features are consistent with direct EPOR agonism, which may circumvent the HIF-2α/hepcidin axis.11,12 However, this interpretation is speculative because hepcidin was not measured in our study, and we cannot exclude that concurrent intravenous iron supplementation (given to 24.5% of patients, almost exclusively in HD) contributed to the observed TSAT elevation. Prospective studies with serial hepcidin measurements are needed to separate drug-mediated iron mobilization from exogenous iron repletion. Because albumin and iron indices improved concurrently, the observed Hb increase should be interpreted as the combined result of erythropoietic stimulation, iron repletion, and improved nutritional status; their relative contributions cannot be separated in this uncontrolled design. These temporal patterns suggest that early reassessment with response-adaptive dosing may be preferable to fixed-interval titration, particularly in patients with severe baseline anemia or elevated inflammatory burden.25

Dialysis vintage of 12–36 months was the strongest predictor of attenuated treatment response. This U-shaped association was not observed when dialysis vintage was modeled as a continuous variable, which likely reflects residual renal function in early dialysis patients and erythropoietic adaptation in long-term dialysis populations. Predictors differed between modalities. BMI was specific to HD (OR 1.38), and this may reflect larger distribution volumes and inflammation-mediated erythropoietic hyporesponsiveness, although the evidence is conflicting. Higher BMI has also been associated with erythropoietin dose-saving in hemodialysis patients.26 Meanwhile, diabetes (OR 3.75) and CVD were specific to PD, reflecting erythropoietic impairment in diabetic nephropathy and potential confounding by cardiorenal syndrome. The positive association between CVD and Hb response in PD likely reflects indication bias: CVD patients start dialysis at lower baseline Hb, leaving more room for improvement. These modality-specific patterns suggest that individualized risk assessment using vintage, baseline Hb, and comorbidities is more useful than modality classification alone. In the month-1 sensitivity analysis, intravenous iron in HD and peritoneal Kt/V in PD were significant early-response predictors, but these effects were not seen at month 3. This suggests iron repletion and dialysis adequacy have the greatest impact during the initial response phase.

The thromboembolic rate of 2.6% (all vascular access-related thromboses), with no systemic thrombotic events, is reassuring and may reflect conservative contemporary Hb targets.5–7 The cardiovascular event rate of 4.7% is consistent with expected frequency in this population. All deaths occurred in patients with pre-existing CVD. The absence of blood pressure elevation may reflect pegmolesatide’s distinct receptor binding.13,14 These safety findings are provisionally supported by a pooled post hoc analysis of two phase 3 trials.27 Preliminary in vitro and animal data suggest that EPOR-CD131-dependent JAK2/STAT3 inhibition and mitochondrial protection may attenuate uremic cardiomyocyte hypertrophy.28,29 Systematic anti-drug antibody testing was not performed; formal immunogenicity surveillance remains essential.30 Whether pegmolesatide’s selective EPOR agonism avoids the systemic risks of HIF pathway activation requires substantiation in head-to-head trials.

Clinical experience with peginesatide acetate (Omontys) underscores the necessity of regular immunogenicity surveillance. This agent was withdrawn from the market in 2013 after reports of severe hypersensitivity reactions, including anaphylaxis in approximately 0.02% of patients, with fatal cases reported at approximately 1 in 2,500 treated patients.31 Most adverse reactions were associated with multi-dose vial formulations and intravenous delivery, and pre-existing anti-PEG antibodies have been identified as a contributing factor.31 Pegmolesatide mitigates these risks through single-dose prefilled syringes, exclusive subcutaneous administration, and a third-generation branched PEG structure distinct from that of peginesatide. No hypersensitivity events were observed in our cohort.

Notably, guideline recommendations diverge on the upper Hb boundary. This cohort was managed according to the Chinese renal anemia guideline,19 which recommends an upper limit of 130 g/L, while the recommendation of the KDIGO 2026 guideline,32 published after our observation period, advises against sustained Hb concentrations above 115 g/L (11.5 g/dL). Consequently, some patients, particularly PD patients in the late treatment phase, maintained Hb concentrations between 115 and 130 g/L. The study was not powered to detect safety differences at these Hb levels. Nevertheless, the sustained late-phase Hb rise observed in PD patients suggests that, if a more conservative ceiling is applied, earlier dose down-titration and individualized upper targets warrant prospective evaluation.

This study has several limitations. First, the observational design without a concurrent control group precludes definitive attribution of the Hb increase to pegmolesatide alone; regression to the mean and the concurrent improvements in iron status and nutrition cannot be excluded. Second, the switching population was heterogeneous: reasons for switching were not mutually exclusive, and while inadequate anemia control was cited in 80.2% of patients, 19.8% already had Hb ≥ 110 g/L at baseline, consistent with non-efficacy reasons for switching; the cohort should therefore not be interpreted as strictly ESA-refractory. Third, month-12 retention was low (31/192, 16.1%); however, decomposition showed this was predominantly administrative censoring: 65 patients were eligible for 12-month follow-up, of whom 31 (47.7%) completed it, and only 5 of the 34 non-completers discontinued for inadequate response; non-completers showed early hemoglobin responses comparable to completers. Month-12 findings should nevertheless be regarded as descriptive. Fourth, the predominantly Asian cohort (mean BMI 23.0 kg/m²) may limit generalizability. Fifth, the non-standardized dosing protocol introduces treatment selection bias, and immortal time bias may inflate apparent safety. Finally, random slope models did not converge in the linear mixed-effects analysis, which assumes homogeneous trajectories and may underestimate between-patient variability.

Conclusion

In this 12-month real-world study of 192 Chinese maintenance dialysis patients switched from prior ESA or HIF-PHI therapy, monthly pegmolesatide was associated with rapid and sustained hemoglobin improvement through month 3 (the pre-specified primary endpoint) and acceptable safety over 12 months of follow-up. The nonlinear trajectory, rapid first-month rise followed by deceleration and stabilization, together with the ceiling effect and the associations of baseline Hb, albumin, and dialysis vintage 12–36 months with response, suggests that response-adaptive individualized dosing deserves prospective evaluation. These findings are limited by the uncontrolled design, the heterogeneous switching population, and substantial month-12 attrition, and require confirmation in randomized controlled trials with concurrent controls and hard clinical endpoints.

Funding Statement

This study was supported by the Traditional Chinese Medicine Science and Technology Plan Project of Zhejiang Province (Grant Nos. 2025ZL506 and 2026ZL0755) and the Natural Science Foundation of Ningbo Municipality (Grant No. 2023J160).

Data Sharing Statement

The datasets generated and/or analyzed during the current study are not publicly available due to institutional data protection policies and patient privacy regulations but are available from the corresponding author on reasonable request, subject to institutional review board approval and data use agreements.

Ethics Approval and Consent to Participate

The study protocol was approved by the Ethics Committee of the First Affiliated Hospital of Ningbo University (approval number: 2026-114) on March 27, 2026, prior to data extraction and analysis. The requirement for individual informed consent was waived by the ethics committee given the retrospective nature of the study, consistent with national regulations for retrospective research using de-identified routinely collected clinical data.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors declare that they have no competing interests.