Endostatin, chronic kidney disease (CKD) and anemia among older people: a secondary analysis of the screening for CKD among older people across Europe (SCOPE) study Authors: Andrea Corsonello, Luca Soraci, Johan Ärnlöv, Axel C. Carlsson, Tobias Rudholm Feldreich, Anders Larsson, Regina Roller-Wirnsberger, Gerhard Wirnsberger, Francesco Mattace-Raso, Lisanne Tap, Francesc Formiga, Rafael Moreno-González, Tomasz Kostka, Agnieszka Guligowska, Ronit Ben-Romano, Itshak Melzer, Christian Weingart, Robert Kob, Cornel C. Sieber, Lucia Muglia, Fabrizia Lattanzio, the SCOPE study investigators Published: 2026 ABSTRACT BackgroundAnemia is a common and debilitating complication of chronic kidney disease (CKD), but its pathogenesis remains incompletely understood. Endostatin, an anti-angiogenic peptide that is elevated in CKD, may impair erythropoiesis through vascular dysfunction. We investigated the relationship between circulating endostatin and both prevalent and incident anemia in older adults, as well as whether kidney function modified this association.MethodsWe analyzed data from 2,008 participants aged ≥ 75 years enrolled in the Screening for CKD among Older People across Europe (SCOPE) prospective cohort. Cross-sectional associations between standardized log-transformed endostatin and hemoglobin levels or prevalent anemia were assessed using linear and logistic regression, respectively. Dose-response relationships were explored across endostatin tertiles. Longitudinal analyses included 1,394 non-anemic individuals followed for two years; incident anemia was assessed using Fine-Gray competing risk models, with death treated as a competing event. Models were progressively adjusted for demographics, comorbidities, kidney function, iron status, medications, and baseline hemoglobin. Sensitivity analyses included Winsorization and subgroup interaction testing.ResultsAt baseline, 405 participants (20.2%) had anemia. Higher endostatin levels were independently associated with lower hemoglobin levels (β -0.21, 95% CI -0.28 to -0.14) and higher odds of prevalent anemia (OR, 95% CI: 1.38, 1.18–1.62). During follow-up, 159 of 1,394 participants (11.4%) developed anemia; higher endostatin levels predicted incident anemia (sHR, 95% CI: 1.40, 1.17–1.69), with more than a twofold higher risk in the highest tertile. Associations were stronger among patients with CKD and were significantly modified by eGFR, advanced age, and the absence of diabetes.ConclusionsThese findings suggest that vascular dysfunction may contribute to anemia in older adults and identify endostatin as a potential biomarker of the risk of anemia, particularly among individuals with CKD.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12967-026-08601-4. Background Chronic kidney disease (CKD) affects approximately 10% of the global population, with current data showing age-adjusted mortality rates of 17.76 per 100,000 people, and is projected to become the fifth leading cause of death by 2040 [1]. Anemia is one of the most prevalent and debilitating comorbidities in CKD, significantly impacting quality of life, increasing the risk of cardiovascular events, and accelerating disease progression [2, 3]. The pathogenesis of anemia in CKD is multifactorial, involving reduced erythropoietin production, iron dysregulation, chronic inflammation, gut dysbiosis, and alterations in kidney function, which collectively affect red blood cell production and lead to decreased hemoglobin levels [4–8]. While current treatment approaches primarily rely on erythropoiesis-stimulating agents (ESAs) and iron supplementation [5], the underlying mechanisms contributing to anemia in CKD are still incompletely understood. Moreover, traditional biomarkers used to assess iron status, such as ferritin and transferrin, are often influenced by CKD-related factors and chronic inflammation, complicating accurate diagnosis. Recent evidence has highlighted the role of endothelial dysfunction in CKD progression [9, 10]. The endothelium plays a critical role in regulating the balance between vascular tone, blood flow, and the availability of erythropoietin, a hormone that stimulates red blood cell production. In CKD, endothelial dysfunction disrupts these processes by reducing nitric oxide availability, increasing vascular resistance, and contributing to systemic inflammation [11], all of which may impair erythropoiesis and exacerbate anemia [10, 12]. This process is further exacerbated by factors such as innate immune responses and the presence of microthrombi, which contribute to sluggish blood flow and endothelial cell damage [10]. One of the mechanisms involved in endothelial dysfunction in CKD is the enhanced generation of endostatin, the antiangiogenic 20-kDa fragment of collagen XVIII [13], which may promote microvascular rarefaction and further vascular and parenchymal pathology [10]. Endostatin concentrations were found to be higher in patients with solid tumors and hematological malignancies, where they correlate with tumor invasion and adverse prognosis [14–16]; furthermore, several studies have demonstrated associations between high plasma endostatin levels and both renal function decline and cardiovascular complications in older patients with CKD [17–21]. However, the relationship between endostatin and anemia in CKD has not yet been investigated. Given this background, we hypothesized that elevated endostatin levels could contribute to the development of anemia in CKD by disrupting the vascular and hematopoietic environments. To test this hypothesis, we aimed to explore the associations between endostatin levels and both prevalent and incident anemia in older people across Europe. We also aimed to evaluate whether these associations were modified by impaired kidney function and other relevant clinical factors. Methods Data source and study design The present study used data from the Screening for Chronic Kidney Disease among Older People across Europe (SCOPE) study (European Grant Agreement no. 636849). The methods of the SCOPE study have been described elsewhere [22]. Briefly, all participants aged 75 years or older attending outpatient services at participating institutions from August 2016 to August 2018 were asked to participate. After providing written informed consent, enrolled participants underwent an extensive assessment including demographic data, socioeconomic status, comprehensive geriatric assessment, assessment of chronic diseases and prescribed medications, and venous blood and urine sampling. Participants were followed up for 24 months, as previously described [23]. The study protocol was approved by the ethics committees at all participating institutions and complied with the Declaration of Helsinki and Good Clinical Practice Guidelines. Sample selection Overall, 2,461 participants were initially enrolled in the study. Of them, 200 were excluded because baseline estimated glomerular filtration rate (eGFR) and albumin-to-creatinine ratio (ACR) data were missing, and 175 participants were excluded because of missing data on plasma endostatin measurement. Among the remaining 2,086 participants, 78 were excluded because of missing baseline hemoglobin and iron status assessments, thus leaving a final sample of 2,008 patients for the primary cross-sectional analysis. The association between endostatin levels and 2-year incident anemia was investigated after excluding 405 patients with anemia at the baseline visit, and 209 participants with no follow-up information, resulting in a subsample of 1,394 patients for the longitudinal analysis. Patients with missing follow-up data were significantly older (mean age, SD 81.2, 4.2 vs. 79.9, 3.8 years; p < 0.001), more commonly women (66% vs. 58.1%, p = 0.036), and had a higher prevalence of hypertension (81.8% vs. 73.8%; p = 0.016), atrial fibrillation (20.1% vs. 12.8%; p = 0.006), congestive heart failure (CHF) (23.9% vs. 13.6%, p < 0.001), coronary artery disease (CAD) (16.7% vs. 10.7%; p = 0.015), and diabetes (26.8% vs. 19.4%; p = 0.017) compared with included participants. They also presented with lower eGFR (median 53.5, IQR 44.0–60.8 vs. 57.2, IQR 48.7–65.0 ml/min/1.73 m²; p < 0.001), higher ACR (median 13.0, IQR 4.2–31.1 vs. 9.9, IQR 3.6–24.9 mg/g; p = 0.032), and higher endostatin levels (median 67.9, IQR 54.4–82.3 vs. 60.8, IQR 49.5–76.4 ng/mL; p < 0.001). Plasma endostatin The collection of clinical and biochemical data has been described in detail elsewhere [22]. Venous blood samples for the measurement of endostatin were drawn at the baseline visit in the morning after an overnight fast and stored at − 80 °C until analysis. Endostatin was measured using a commercial sandwich ELISA (DY1098, R&D Systems, Minneapolis, MN, USA) with a detection threshold of 0.063 ng/mL. A monoclonal antibody specific for endostatin was coated onto microtiter plates. The plates were blocked with bovine serum albumin and then washed. Samples and standards were added to the wells, allowing the peptide to bind to the immobilized antibodies. A biotinylated endostatin-specific antibody was added after washing. A streptavidin–HRP conjugate was added to the wells after incubation and washing, and a substrate solution was added after another incubation and washing cycle. The absorbance was measured using a SpectraMax 250 (Molecular Devices, Sunnyvale, CA, USA). Endostatin values were determined by comparing the optical density of the samples with the standard curve. The assay was calibrated against highly purified recombinant human endostatin. Measurements were performed blinded, without knowledge of the clinical information. The total coefficient of variation for the endostatin assay was approximately 6%. Study outcomes We separately examined the association between serum endostatin levels and the prevalence of anemia at baseline and between serum endostatin levels and the incidence of anemia during the two-year follow-up. Anemia was defined as the presence of a hemoglobin concentration < 12 g/dL in women and < 13 g/dL in men [24, 25]. Covariates Venous blood samples for the measurement of traditional metabolites were collected at all in-person study visits (baseline, 1-year, and 2-year study visits). Iron status was assessed by using age-adapted definitions to increase diagnostic sensitivity in older individuals, based on serum ferritin and transferrin saturation (TSAT) values; abnormal iron tests were defined as serum ferritin ≤ 100 µg/L, indicating absolute iron deficiency, or serum ferritin < 300 µg/L with TSAT ≤ 20%, indicating functional iron deficiency [26, 27]. Nutritional deficiencies were also considered, including vitamin B12 deficiency, defined as a serum vitamin B12 level < 150 pg/mL, and folic acid deficiency, defined as a folic acid level ≤ 5 ng/mL [28]. Baseline serum creatinine was measured using an isotope dilution mass spectrometry(IDMS)-traceable standard method; eGFR was calculated using the creatinine-based Berlin Initiative Study (BIS) equation, which was specifically developed in a population older than 70 years [29]. The presence of CKD at the baseline visit was defined as having eGFR < 60 ml/min/1.73 m2 according to the Kidney Disease: Improving Global Outcomes (KDIGO) classification [30]. Other covariates included demographics (age, sex), history of chronic diseases including CHF, CAD, atrial fibrillation, osteoporosis, diabetes, cancer, chronic obstructive pulmonary disease (COPD), and stroke. Prescribed medications coded according to the ATC classification were also investigated, and drugs potentially associated with anemia were included: proton pump inhibitors (PPIs), antithrombotics, antiplatelet medications, corticosteroids, and antianemic medications including iron supplements, vitamin B12/folic acid, and erythropoietin (EPO). Statistical analysis Descriptive analyses were first conducted to compare patients with and without anemia at baseline in terms of demographic, clinical, pharmacological, and laboratory characteristics. Continuous variables were reported as means and standard deviations (SDs) or medians and interquartile ranges (IQRs), according to their distribution, which was evaluated by visual inspection and the Shapiro-Wilk test. Categorical variables were reported as numbers and percentages (%). The two-sample t-test and Mann–Whitney U test were used to compare continuous variables, while Pearson’s Chi-square test was used to compare categorical variables. In all analyses, serum endostatin levels were analyzed as either a continuous variable or as tertiles. Continuous endostatin was logarithmically transformed and then standardized because of its skewed distribution. Cross-sectional analysis The correlation between serum endostatin and baseline hemoglobin or eGFR was first investigated using Spearman’s rank correlation coefficient. Bivariate and multivariate models were then constructed to evaluate the relationship between endostatin levels and (i) hemoglobin concentrations, using linear regression; and (ii) the presence of anemia, using logistic regression. Three progressive models were considered: Model A, adjusted for age and sex; Model B, adjusted for age, sex, hypertension, diabetes, CHF, CAD, atrial fibrillation, osteoporosis, stroke, COPD, cancer, CKD, ACR ≥ 30 mg/g, and iron deficiency; Model C, adjusted for age, sex, hypertension, diabetes, CHF, CAD, atrial fibrillation, osteoporosis, stroke, COPD, cancer, CKD, ACR ≥ 30 mg/g, iron deficiency, PPIs, antianemic medications, antithrombotics, and corticosteroids. Longitudinal analysis The association between baseline endostatin levels and incident anemia was explored in the subsample of 1,394 patients without anemia at the baseline visit. Incident anemia was ascertained at scheduled follow-up visits at one and two years after the baseline visit, when hemoglobin concentration was remeasured. Participants were classified as having developed incident anemia if their hemoglobin concentration at any follow-up visit was below the WHO thresholds (< 13.0 g/dL in men and < 12.0 g/dL in women) in the absence of prevalent anemia at baseline. The time to event was defined as the number of days from the baseline visit to the date of the earliest follow-up visit at which anemia was first ascertained (either the one-year or two-year visit). For participants who did not develop anemia at either follow-up visit, follow-up time was censored at the date of the last available visit. Because death during follow-up could preclude the observation of incident anemia, time-to-event analyses were performed within a competing-risk framework using Fine–Gray subdistribution hazard models, treating death as a competing event. More specifically, the outcome was operationalized as follows: incident anemia represented the event of interest, while death before or at the follow-up visit without incident anemia was treated as the competing event; event-free survival, defined as being alive without anemia at the last available follow-up visit, represented the reference status. Subdistribution hazard ratios (sHRs) and corresponding 95% confidence intervals (CIs) were estimated for the association between standardized log-transformed endostatin levels and incident anemia. Analysis was repeated by using endostatin tertiles instead of the continuous variable. Multicollinearity between covariates was evaluated using the variance inflation factor (VIF). Multivariable adjustment followed the same rules used for the logistic and linear regression models, with the addition of Model D, which also included baseline hemoglobin values. Stratified analyses with interaction terms were also carried out by repeating cross-sectional multivariable logistic regression analysis for prevalent anemia and longitudinal models for incident anemia after stratifying by relevant covariates included in the study. Some variables, including antianemic medications and corticosteroids, were excluded from stratification because of the low number of events, which prevented the calculation of reliable odds estimates. To assess the robustness of study findings, we performed sensitivity analyses by Winsorizing extreme endostatin values at the 5th and 99th percentiles to evaluate the influence of outliers on effect estimates. All statistical analyses were conducted using R software version 4.2. All statistical tests were two-sided, and a p value < 0.05 was considered statistically significant. Results Main characteristics of the study population The characteristics of the study population stratified by anemia status are shown in Table 1. Among the 2,008 patients included in the study, 405 (20.2%) had a diagnosis of anemia at the baseline visit. Patients with anemia were older and more frequently male than those without anemia. All comorbid conditions and medications included in the analysis, as well as iron deficiency, were more prevalent in patients with anemia, while no significant difference was observed in vitamin B12 or folate status. Furthermore, patients with anemia had lower eGFR and higher circulating endostatin levels than patients without anemia. Table 1General characteristics of study subjects stratified according to the presence or absence of anemia at the baselineAll (n = 2,008)No anemia (n = 1,603)Anemia (n = 405)P valueAge, years, mean (SD)80.4 (4.1)80.0 (3.9)81.8 (4.7)< 0.001Male sex, n (%)880 (43.8)655 (40.9)225 (55.6)< 0.001 Chronic diseases Hypertension, n (%)1,537 (76.5)1,200 (74.9)337 (83.2)< 0.001Atrial fibrillation, n (%)310 (15.4)221 (13.8)89 (22.0)< 0.001CHF, n (%)337 (16.8)239 (14.9)98 (24.2)< 0.001CAD, n (%)268 (13.3)184 (11.5)84 (20.7)< 0.001Diabetes, n (%)482 (24.0)326 (20.3)156 (38.5)< 0.001Cancer, n (%)346 (17.2)262 (16.3)84 (20.7)0.043COPD, n (%)250 (12.5)181 (11.3)69 (17.0)0.002Osteoporosis, n (%)602 (30.0)462 (28.8)140 (34.6)0.028Stroke, n (%)114 (5.7)83 (5.2)31 (7.7)0.071 Laboratory values eGFR, ml/min/1.73 m2, median (IQR)54.8 (44.1–63.5)56.5 (48.1–64.5)43.0 (32.7–54.9)< 0.001eGFR category, n (%)< 0.001 ≥ 60677 (33.7)608 (37.9)69 (17.0) 45-59.9795 (39.6)679 (42.4)116 (28.6) < 45536 (26.7)316 (19.7)220 (54.3)ACR, mg/g, median (IQR)11.4 (3.9–31.3)10.2 (3.6–25.9)22.5 (6.0-91.2)< 0.001ACR category, n (%)< 0.001 < 301,481 (73.7)1,256 (78.3)225 (55.5) ≥ 30527 (26.2)347 (21.6)180 (44.4)Absolute iron deficiency, n (%)965 (48.1)749 (46.7)216 (53.3)0.020Functional iron deficiency, n (%)80 (4.0)56 (3.5)24 (5.9)0.036Low vitamin B12 and/or folate, n (%)425 (24.0)344 (23.9)81 (24.7)0.813Endostatin, ng/mL, median (IQR)64.3 (51.4–83.1)61.5 (50.1–77.1)80.2 (60.6-112.3)< 0.001 Medications PPI, n (%)705 (35.1)515 (32.1)190 (46.9)< 0.001Corticosteroids, n (%)74 (3.7)48 (3.0)26 (6.4)0.002Antithrombotics, n (%)1,016 (50.6)752 (46.9)264 (65.2)< 0.001Antianemics, n (%)162 (8.1)92 (5.7)70 (17.3)< 0.001Notes. ACR, albumin-to-creatinine ratio; CAD, coronary artery disease; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; eGFR, estimated glomerular filtration rate; PPI, proton pump inhibitors Table 1General characteristics of study subjects stratified according to the presence or absence of anemia at the baselineAll (n = 2,008)No anemia (n = 1,603)Anemia (n = 405)P valueAge, years, mean (SD)80.4 (4.1)80.0 (3.9)81.8 (4.7)< 0.001Male sex, n (%)880 (43.8)655 (40.9)225 (55.6)< 0.001 Chronic diseases Hypertension, n (%)1,537 (76.5)1,200 (74.9)337 (83.2)< 0.001Atrial fibrillation, n (%)310 (15.4)221 (13.8)89 (22.0)< 0.001CHF, n (%)337 (16.8)239 (14.9)98 (24.2)< 0.001CAD, n (%)268 (13.3)184 (11.5)84 (20.7)< 0.001Diabetes, n (%)482 (24.0)326 (20.3)156 (38.5)< 0.001Cancer, n (%)346 (17.2)262 (16.3)84 (20.7)0.043COPD, n (%)250 (12.5)181 (11.3)69 (17.0)0.002Osteoporosis, n (%)602 (30.0)462 (28.8)140 (34.6)0.028Stroke, n (%)114 (5.7)83 (5.2)31 (7.7)0.071 Laboratory values eGFR, ml/min/1.73 m2, median (IQR)54.8 (44.1–63.5)56.5 (48.1–64.5)43.0 (32.7–54.9)< 0.001eGFR category, n (%)< 0.001 ≥ 60677 (33.7)608 (37.9)69 (17.0) 45-59.9795 (39.6)679 (42.4)116 (28.6) < 45536 (26.7)316 (19.7)220 (54.3)ACR, mg/g, median (IQR)11.4 (3.9–31.3)10.2 (3.6–25.9)22.5 (6.0-91.2)< 0.001ACR category, n (%)< 0.001 < 301,481 (73.7)1,256 (78.3)225 (55.5) ≥ 30527 (26.2)347 (21.6)180 (44.4)Absolute iron deficiency, n (%)965 (48.1)749 (46.7)216 (53.3)0.020Functional iron deficiency, n (%)80 (4.0)56 (3.5)24 (5.9)0.036Low vitamin B12 and/or folate, n (%)425 (24.0)344 (23.9)81 (24.7)0.813Endostatin, ng/mL, median (IQR)64.3 (51.4–83.1)61.5 (50.1–77.1)80.2 (60.6-112.3)< 0.001 Medications PPI, n (%)705 (35.1)515 (32.1)190 (46.9)< 0.001Corticosteroids, n (%)74 (3.7)48 (3.0)26 (6.4)0.002Antithrombotics, n (%)1,016 (50.6)752 (46.9)264 (65.2)< 0.001Antianemics, n (%)162 (8.1)92 (5.7)70 (17.3)< 0.001Notes. ACR, albumin-to-creatinine ratio; CAD, coronary artery disease; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; eGFR, estimated glomerular filtration rate; PPI, proton pump inhibitors General characteristics of study subjects stratified according to the presence or absence of anemia at the baseline Notes. ACR, albumin-to-creatinine ratio; CAD, coronary artery disease; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; eGFR, estimated glomerular filtration rate; PPI, proton pump inhibitors Median (IQR) endostatin values varied according to the presence or absence of specific clinical conditions (Table S1). Plasma endostatin levels were significantly higher in patients aged > 80 years, as well as in those with hypertension, diabetes, CHF, CAD, atrial fibrillation, osteoporosis, stroke, cancer, more deteriorated eGFR and ACR; they were also higher in patients taking PPIs, antithrombotics, and antianemic medications. Cross-sectional association between endostatin levels and baseline hemoglobin values or anemia prevalence Plasma endostatin was negatively correlated with hemoglobin and eGFR at baseline in the overall study population (Fig. 1, upper panels). Stratification by CKD status showed that the negative correlation between endostatin and hemoglobin was present only among patients with CKD (Fig. 1, lower panels). Fig. 1Correlations of baseline endostatin with hemoglobin or eGFR values (upper panels) and with hemoglobin among patients with or without CKD (lower panels). Scatter plots showing Spearman’s correlation between circulating endostatin levels (ng/mL) and hemoglobin (g/dL) (left panels) and eGFR (mL/min/1.73 m²) (right panels) in the whole study population (upper panels, n = 2,008) and stratified by CKD status (lower panels: no CKD, n = 677; CKD, n = 1,331). CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate Correlations of baseline endostatin with hemoglobin or eGFR values (upper panels) and with hemoglobin among patients with or without CKD (lower panels). Scatter plots showing Spearman’s correlation between circulating endostatin levels (ng/mL) and hemoglobin (g/dL) (left panels) and eGFR (mL/min/1.73 m²) (right panels) in the whole study population (upper panels, n = 2,008) and stratified by CKD status (lower panels: no CKD, n = 677; CKD, n = 1,331). CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate Linear regression models exploring the association between endostatin levels and hemoglobin confirmed these preliminary findings (Table S2): both standardized log-transformed endostatin and increasing endostatin tertiles were associated with lower hemoglobin values. Stratification by CKD status showed that this association was significant only in patients with CKD (Table S2). Similar findings were obtained when prevalent anemia was considered as the outcome in multivariable logistic regression models (Table S3); furthermore, the association between endostatin and prevalent anemia was consistent across strata of covariates, but some variables significantly modified the association. Specifically, the association was significantly modified by CKD status (OR 1.43, 95% CI 1.18–1.74 in CKD vs. OR 1.16, 95%CI 0.84–1.61 in non-CKD; P for interaction = 0.002), by hypertension (OR 1.45, 95% CI 1.21–1.74 in hypertensive vs. OR 1.17, 95%CI 0.82–1.67 in non-hypertensive patients; P for interaction = 0.008), and CHF (OR 1.78, 95% CI 1.19–2.73 in CHF vs. OR 1.31, 95% CI 1.09–1.56 in non-CHF; P for interaction = 0.017) (Table S4, Fig. 2). Fig. 2Stratified logistic regression analysis of baseline association between log-transformed endostatin and prevalent anemia. Forest plot showing the cross-sectional association between standardized log-transformed endostatin levels and prevalent anemia across prespecified subgroups (Model C, fully adjusted for age, sex, comorbidities, kidney function, iron status, and medications). Each point represents the odds ratio (OR) per one standard deviation increase in log-transformed endostatin, with horizontal bars indicating 95% confidence intervals. The red dashed vertical line corresponds to the null value (OR = 1.0). The right-hand columns display the subgroup-specific OR (95% CI) and the P value for multiplicative interaction for each subgroup pair. Statistically significant P values for interaction (P < 0.05) are shown in bold red. Asterisks (*) denote subgroups with a statistically significant interaction: hypertension (P for interaction = 0.008), congestive heart failure (P for interaction = 0.017), and chronic kidney disease (P for interaction = 0.002). Full results for all subgroups are reported in Supplementary Table 4. ACR, albumin-to-creatinine ratio; CAD, coronary artery disease; CHF, congestive heart failure; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; OR, odds ratio; PPIs, proton pump inhibitors Stratified logistic regression analysis of baseline association between log-transformed endostatin and prevalent anemia. Forest plot showing the cross-sectional association between standardized log-transformed endostatin levels and prevalent anemia across prespecified subgroups (Model C, fully adjusted for age, sex, comorbidities, kidney function, iron status, and medications). Each point represents the odds ratio (OR) per one standard deviation increase in log-transformed endostatin, with horizontal bars indicating 95% confidence intervals. The red dashed vertical line corresponds to the null value (OR = 1.0). The right-hand columns display the subgroup-specific OR (95% CI) and the P value for multiplicative interaction for each subgroup pair. Statistically significant P values for interaction (P < 0.05) are shown in bold red. Asterisks (*) denote subgroups with a statistically significant interaction: hypertension (P for interaction = 0.008), congestive heart failure (P for interaction = 0.017), and chronic kidney disease (P for interaction = 0.002). Full results for all subgroups are reported in Supplementary Table 4. ACR, albumin-to-creatinine ratio; CAD, coronary artery disease; CHF, congestive heart failure; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; OR, odds ratio; PPIs, proton pump inhibitors Longitudinal association between endostatin levels and incident anemia risk Among the 1,394 patients included in the longitudinal analysis who were free of anemia at baseline, 159 (11.4%) developed incident anemia during a median (IQR) follow-up of 23 (22–24) months. During the same period, 17 participants (1.2%) died before anemia assessment and were therefore treated as competing events in Fine–Gray subdistribution hazard models. The characteristics of patients without anemia at baseline, stratified by endostatin tertiles, are reported in Table 2. Increasing endostatin levels were associated with older age, a higher prevalence of hypertension, cardiovascular diseases, diabetes, cancer, and iron deficiency, and a lower prevalence of osteoporosis. The highest endostatin tertile was also characterized by lower eGFR, higher ACR, lower use of PPIs, and higher use of antianemic medications. Table 2Characteristics of patients without anemia at the baseline included in longitudinal analysis and stratified by END tertilesOverall (n = 1,394)pEND T1 (n = 465)pEND T2 (n = 464)pEND T3 (n = 465)p valueEndostatin, ng/mL, range15.6–22215.6–53.353.4–69.469.5–222-Age, years, mean (SD)79.9 (3.8)79.5 (3.5)79.6 (3.7)80.4 (4.1)< 0.001Male sex, n (%)584 (41.9)200 (43.3)183 (39.4)201 (43.2)0.422Hypertension, n (%)1029 (73.8)324 (69.7)343 (73.9)362 (77.8)0.018Atrial fibrillation, n (%)179 (12.8)54 (11.6)51 (11.0)74 (15.9)0.051CHF, n (%)189 (13.6)58 (12.5)53 (11.4)78 (16.8)0.041CAD, n (%)149 (10.7)34 (7.3)54 (11.6)61 (13.1)0.012Diabetes, n (%)270 (19.4)73 (15.7)85 (18.3)112 (24.1)0.004Cancer, n (%)229 (16.4)65 (14.0)71 (15.3)93 (20.0)0.034COPD, n (%)159 (11.4)61 (13.1)46 (9.9)52 (11.2)0.302Osteoporosis, n (%)391 (28.0)146 (31.4)136 (29.3)109 (23.4)0.020Stroke, n (%)66 (4.7)14 (3.0)24 (5.2)28 (6.0)0.083eGFR, ml/min/1.73 m2, mean (SD)56.5 (12.9)62.6 (10.2)57.7 (10.7)49.3 (13.7)< 0.001eGFR category, n (%)< 0.001 ≥ 60551 (39.5)268 (57.6)186 (40.1)97 (20.9) 45-59.9585 (42.0)180 (38.7)220 (47.4)185 (39.8) < 45258 (18.5)17 (3.7)58 (12.5)183 (39.4)20% eGFR decline, n (%)83 (6.0)28 (6.0)24 (5.2)31 (6.7)0.461ACR, mg/g, median (IQR)9.9 (3.6–24.9)7.0 (2.8–16.3)10.2 (3.7–23.5)13.0 (4.7–38.0)< 0.001ACR category, n (%)< 0.001 < 301,106 (79.3)402 (86.4)376 (81.0)328 (70.5) ≥ 30288 (20.7)63 (13.5)88 (19.0)137 (29.5)Absolute iron deficiency, n (%)651 (46.7)237 (51.0)195 (42.0)219 (47.1)0.023Functional iron deficiency, n (%)47 (3.4)17 (3.7)11 (2.4)19 (4.1)0.164Low vitamin B12 and/or folate, n (%)297 (21.3)106 (22.8)95 (20.5)96 (20.6)0.109PPI, n (%)436 (31.3)170 (36.6)143 (30.8)123 (26.5)0.004Antithrombotics, n (%)634 (45.5)199 (42.8)209 (45.0)226 (48.6)0.200Antianemics, n (%)86 (6.2)24 (5.2)22 (4.7)40 (8.6)0.027Corticosteroids, n (%)40 (2.9)16 (3.4)8 (1.7)16 (3.4)0.195Notes. ACR, albumin-to-creatinine ratio; CAD, coronary artery disease; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; eGFR, estimated glomerular filtration rate; PPI, proton pump inhibitors Table 2Characteristics of patients without anemia at the baseline included in longitudinal analysis and stratified by END tertilesOverall (n = 1,394)pEND T1 (n = 465)pEND T2 (n = 464)pEND T3 (n = 465)p valueEndostatin, ng/mL, range15.6–22215.6–53.353.4–69.469.5–222-Age, years, mean (SD)79.9 (3.8)79.5 (3.5)79.6 (3.7)80.4 (4.1)< 0.001Male sex, n (%)584 (41.9)200 (43.3)183 (39.4)201 (43.2)0.422Hypertension, n (%)1029 (73.8)324 (69.7)343 (73.9)362 (77.8)0.018Atrial fibrillation, n (%)179 (12.8)54 (11.6)51 (11.0)74 (15.9)0.051CHF, n (%)189 (13.6)58 (12.5)53 (11.4)78 (16.8)0.041CAD, n (%)149 (10.7)34 (7.3)54 (11.6)61 (13.1)0.012Diabetes, n (%)270 (19.4)73 (15.7)85 (18.3)112 (24.1)0.004Cancer, n (%)229 (16.4)65 (14.0)71 (15.3)93 (20.0)0.034COPD, n (%)159 (11.4)61 (13.1)46 (9.9)52 (11.2)0.302Osteoporosis, n (%)391 (28.0)146 (31.4)136 (29.3)109 (23.4)0.020Stroke, n (%)66 (4.7)14 (3.0)24 (5.2)28 (6.0)0.083eGFR, ml/min/1.73 m2, mean (SD)56.5 (12.9)62.6 (10.2)57.7 (10.7)49.3 (13.7)< 0.001eGFR category, n (%)< 0.001 ≥ 60551 (39.5)268 (57.6)186 (40.1)97 (20.9) 45-59.9585 (42.0)180 (38.7)220 (47.4)185 (39.8) < 45258 (18.5)17 (3.7)58 (12.5)183 (39.4)20% eGFR decline, n (%)83 (6.0)28 (6.0)24 (5.2)31 (6.7)0.461ACR, mg/g, median (IQR)9.9 (3.6–24.9)7.0 (2.8–16.3)10.2 (3.7–23.5)13.0 (4.7–38.0)< 0.001ACR category, n (%)< 0.001 < 301,106 (79.3)402 (86.4)376 (81.0)328 (70.5) ≥ 30288 (20.7)63 (13.5)88 (19.0)137 (29.5)Absolute iron deficiency, n (%)651 (46.7)237 (51.0)195 (42.0)219 (47.1)0.023Functional iron deficiency, n (%)47 (3.4)17 (3.7)11 (2.4)19 (4.1)0.164Low vitamin B12 and/or folate, n (%)297 (21.3)106 (22.8)95 (20.5)96 (20.6)0.109PPI, n (%)436 (31.3)170 (36.6)143 (30.8)123 (26.5)0.004Antithrombotics, n (%)634 (45.5)199 (42.8)209 (45.0)226 (48.6)0.200Antianemics, n (%)86 (6.2)24 (5.2)22 (4.7)40 (8.6)0.027Corticosteroids, n (%)40 (2.9)16 (3.4)8 (1.7)16 (3.4)0.195Notes. ACR, albumin-to-creatinine ratio; CAD, coronary artery disease; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; eGFR, estimated glomerular filtration rate; PPI, proton pump inhibitors Characteristics of patients without anemia at the baseline included in longitudinal analysis and stratified by END tertiles Notes. ACR, albumin-to-creatinine ratio; CAD, coronary artery disease; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; eGFR, estimated glomerular filtration rate; PPI, proton pump inhibitors When included in Fine–Gray subdistribution hazard models, both standardized log-transformed endostatin and higher endostatin tertiles were significantly associated with incident anemia (Table 3), even after adjustment for multiple confounders and after further adjustment for baseline hemoglobin levels in model D. Analyses stratified by CKD status revealed that the associations between continuous endostatin levels and incident anemia were stronger among patients with CKD than among those without CKD. A significant interaction was observed between continuous endostatin and baseline eGFR (sHR, 95% CI: 0.98, 0.97–0.99); stratified analyses also showed that the association between endostatin and incident anemia was modified by eGFR (sHR 2.43, 95% CI 1.68–3.51 for eGFR < 45 ml/min/1.73 m² vs. sHR 1.36, 95% CI 1.05–1.75 for eGFR ≥ 45 ml/min/1.73 m²; P for interaction = 0.035); other significant interactions were found with age (sHR 1.61, 95% CI 1.21–2.14 for age > 80 years vs. sHR 1.29, 95% CI 1.01–1.65 for age ≤ 80 years; P for interaction = 0.041), absence of iron deficiency (sHR 1.76, 95% CI 1.33–2.33 in those without iron deficiency vs. sHR 1.16, 95% CI 0.89–1.53 in those with iron deficiency; P for interaction = 0.049), and absence of diabetes (sHR 1.58, 95%CI 1.27–1.96 in those without diabetes vs. sHR 1.11, 95% CI 0.83–1.49 in those with diabetes; P for interaction = 0.021) (Table S5). Table 3Fine-Gray competing risk models showing the association between log endostatin levels and incident anemiaAll patientsModel A, sHR (95%CI)Model B, sHR (95%CI)Model C, sHR (95%CI)Model D, sHR (95%CI) Log Endostatin Log endostatin1.55 (1.30–1.85)1.47 (1.21–1.79)1.55 (1.29–1.87)1.40 (1.17–1.69) Endostatin tertiles Endostatin T1----Endostatin T21.16 (0.74–1.80)1.17 (0.75–1.82)1.28 (0.82–1.99)1.28 (0.81–2.03)Endostatin T32.35 (1.58–3.50)2.06 (1.34–3.17)2.43 (1.60–3.69)2.28 (1.44–3.63)Endostatin*eGFR0.94 (0.93–0.97)*0.95 (0.93–0.97)*0.95 (0.93–0.97)*0.97 (0.95–0.98)* Patients without CKD Log Endostatin Log endostatin1.10 (0.73–1.64)1.14 (0.68–1.67)1.23 (0.83–1.82)1.38 (0.92–2.06) Endostatin tertiles Endostatin T1----Endostatin T21.02 (0.55–1.91)1.15 (0.60–2.21)1.17 (0.60–2.28)1.25 (0.64–1.42)Endostatin T31.71 (0.86–3.43)1.90 (0.94–3.84)2.42 (1.14–5.14)3.27 (1.33–8.04) Patients with CKD Log Endostatin Log endostatin1.72 (1.40–2.10)1.57 (1.26–1.95)1.64 (1.33–2.03)1.43 (1.16–1.75) Endostatin tertiles Endostatin T1----Endostatin T21.31 (0.67–2.53)1.23 (0.63–2.40)1.34 (0.70–2.58)1.20 (0.62–2.32)Endostatin T32.62 (1.44–4.76)2.05 (1.12–3.75)2.51 (1.41–4.49)1.97 (1.10–3.53)Notes. CKD = chronic kidney disease; eGFR = estimated glomerular filtration rate; T = tertile; *p value < 0.05 Table 3Fine-Gray competing risk models showing the association between log endostatin levels and incident anemiaAll patientsModel A, sHR (95%CI)Model B, sHR (95%CI)Model C, sHR (95%CI)Model D, sHR (95%CI) Log Endostatin Log endostatin1.55 (1.30–1.85)1.47 (1.21–1.79)1.55 (1.29–1.87)1.40 (1.17–1.69) Endostatin tertiles Endostatin T1----Endostatin T21.16 (0.74–1.80)1.17 (0.75–1.82)1.28 (0.82–1.99)1.28 (0.81–2.03)Endostatin T32.35 (1.58–3.50)2.06 (1.34–3.17)2.43 (1.60–3.69)2.28 (1.44–3.63)Endostatin*eGFR0.94 (0.93–0.97)*0.95 (0.93–0.97)*0.95 (0.93–0.97)*0.97 (0.95–0.98)* Patients without CKD Log Endostatin Log endostatin1.10 (0.73–1.64)1.14 (0.68–1.67)1.23 (0.83–1.82)1.38 (0.92–2.06) Endostatin tertiles Endostatin T1----Endostatin T21.02 (0.55–1.91)1.15 (0.60–2.21)1.17 (0.60–2.28)1.25 (0.64–1.42)Endostatin T31.71 (0.86–3.43)1.90 (0.94–3.84)2.42 (1.14–5.14)3.27 (1.33–8.04) Patients with CKD Log Endostatin Log endostatin1.72 (1.40–2.10)1.57 (1.26–1.95)1.64 (1.33–2.03)1.43 (1.16–1.75) Endostatin tertiles Endostatin T1----Endostatin T21.31 (0.67–2.53)1.23 (0.63–2.40)1.34 (0.70–2.58)1.20 (0.62–2.32)Endostatin T32.62 (1.44–4.76)2.05 (1.12–3.75)2.51 (1.41–4.49)1.97 (1.10–3.53)Notes. CKD = chronic kidney disease; eGFR = estimated glomerular filtration rate; T = tertile; *p value < 0.05 Fine-Gray competing risk models showing the association between log endostatin levels and incident anemia Notes. CKD = chronic kidney disease; eGFR = estimated glomerular filtration rate; T = tertile; *p value < 0.05 Further stratified analyses (Table S5) showed a significant interaction between endostatin levels and eGFR < 45 ml/min (sHR 1.70, 1.09–2.64, p for interaction = 0.035), with the association between endostatin and anemia becoming stronger as kidney function deteriorated; furthermore, the effect of endostatin on the incidence of anemia was stronger in patients with preserved iron stores (p = 0.049), those without diabetes (p = 0.021) and patients aged > 80 years (p = 0.041). Study results were confirmed in sensitivity analyses after winsorization of extreme endostatin values (Table S6). Discussion The present study provides novel evidence that elevated circulating endostatin levels are independently associated with both lower hemoglobin concentrations and increased prevalence and incidence of anemia in older adults, independent of demographic factors, comorbidities, and medications. Given the putative biological role of endostatin, these results implicate that endothelial dysfunction, triggered by the anti-angiogenic activity of endostatin, may contribute to the pathogenesis of anemia. The development and progression of anemia in CKD are conventionally attributed to the interplay of three main converging mechanisms: insufficient EPO synthesis due to decreased renal mass, hepcidin-mediated functional iron deficiency, and chronic inflammation which blunts proliferation of erythroid progenitors [5]. Despite these pathways remain the primary therapeutic targets, a substantial proportion of patients with CKD continue to suffer from anemia despite correction of EPO deficiency and iron stores, suggesting the existence of other mechanisms that may contribute to the residual burden of the disease [31, 32]. Accordingly, new biomarkers may be useful for improving the diagnosis and management of anemia in CKD. In the cross-sectional analysis, higher endostatin levels were significantly associated with lower hemoglobin concentrations, and this relationship was significant especially among individuals with CKD and hypertension. The longitudinal analysis strengthened these observations and revealed that participants with elevated baseline endostatin values had a substantially higher risk of developing anemia over a two-year follow-up, independent of baseline hemoglobin, kidney function decline, iron deficiency, and comorbid conditions. However, the longitudinal association was significant only among patients with decreased eGFR, and a multiplicative interaction was found between continuous endostatin levels and either eGFR or eGFR < 45 mL/min/1.73 m2. Taken together, these findings indicate that endostatin exerts its strongest clinical impact in older adults with CKD, and that renal microvascular damage and advanced age further enhance its contribution to the development of anemia. Mechanistically, endostatin may contribute to anemia by inhibiting vascular endothelial growth factor (VEGF), a key regulator of EPO production under hypoxic conditions. VEGF supports both the survival of peritubular capillaries and the activation of EPO-producing stromal cells. Experimental models have demonstrated that VEGF inhibition leads to capillary rarefaction and suppression of endogenous EPO synthesis, both of which could be mirrored by increased endostatin levels [33–35]. The imbalance between insufficient pro-angiogenic VEGF and increased anti-angiogenic endostatin has shown to impair angiogenesis in aging kidneys and mediate some of the effects of aging on vascular remodeling [36–39]. Furthermore, plasma endostatin may directly and indirectly inhibit VEGF function [38], which can induce EPO production [33] and may thus play a role in the onset and progression of anemia in CKD. Crucially, this vascular pathway is mechanistically different from but potentially complementary to the classical mechanisms related to iron and EPO deficiency: in CKD, hepcidin excess restricts iron availability for erythropoiesis independently of total body iron stores and EPO deficiency restricts the response of erythroid progenitors [4]. Endostatin may exacerbate both deficiencies by impairing the peritubular microvascular niche on which interstitial cells that produce EPO depend [40, 41]. Moreover, the chronic inflammatory milieu of CKD, which amplifies hepcidin production and suppresses erythroid differentiation, may itself promote collagen turnover and endostatin release [21, 42]; this can create a vicious cycle in which inflammation drives endostatin elevation, which in turn exacerbates vascular rarefaction and EPO insufficiency independently of iron status. The stronger association between endostatin and anemia in hypertensive patients may be mechanistically explained by systemic microvascular rarefaction, which is driven by endothelial progenitor cell dysfunction and impaired angiogenesis in the setting of hypertension [43]. Consequently, elevated endostatin may act synergistically with pre-existing hypertension-related microvascular injury to further suppress the vascular environment required for erythropoietin production, so that the anti-angiogenic effect of endostatin becomes clinically relevant only when superimposed on an already compromised microvascular bed. These pathophysiological insights may have important therapeutic implications. First, the efficacy of traditional ESAs may be limited in individuals with elevated endostatin, who may have reduced vascular capacity to respond to exogenous EPO. ESA resistance is common in CKD and is often attributed to inflammation or iron deficiency, but vascular dysfunction may represent an additional mechanism underlying a poor hematologic response [44]. In this framework, screening for endostatin levels could help stratify patients with ESA-resistant anemia according to the dominant pathophysiological pathway, thus allowing a tailored therapeutic approach. Future studies should address whether monitoring endostatin levels may help identify patients who are less likely to respond to ESA therapy and could benefit from alternative strategies. Second, hypoxia-inducible factor prolyl hydroxylase inhibitors (HIF-PHIs) represent a promising approach to addressing anemia in this context. These agents increase endogenous EPO production by stabilizing HIF transcription factors and up-regulate VEGF and pro-angiogenic pathways [5]. Preclinical studies have shown that HIF signaling may reduce matrix metalloproteinase activity and may thereby decrease the generation of endostatin [45], suggesting that HIF-PHIs could both enhance erythropoiesis and partially counteract the vascular inhibition driven by endostatin. In addition, Guo et al. demonstrated in endothelial cells that endostatin translocated into the nucleus can actively suppress HIF-1α activation by disrupting intracellular zinc(II) homeostasis, leading to reduced HIF-1α mRNA and protein expression [46]. This may create a vicious cycle in which high endostatin levels not only impede VEGF and EPO pathways but may also blunt the response to hypoxia via HIF-1α downregulation. Such dual interference may synergistically impair both erythropoietic drive and vascular adaptation in CKD. Third, the findings of the present study support the emerging concept of “vascular anemia,” a phenotype characterized by impaired endothelial and microvascular support for red blood cell production. This model, originally proposed in chronic heart failure [47], is now increasingly recognized in CKD, where systemic endothelial injury and capillary rarefaction may contribute to hematologic disturbances [48]. Finally, our findings also suggest that the association between endostatin and anemia is modulated by iron status. Specifically, elevated endostatin levels were associated with incident anemia only among individuals without iron deficiency, indicating that endostatin may play a more prominent role when iron availability is adequate, and iron deficiency is not the primary driver of impaired erythropoiesis. This finding is consistent with a hierarchical model of anemia pathogenesis in CKD, in which absolute iron deficiency dominates erythropoietic output when present, whereas vascular mechanisms, including endostatin-mediated EPO suppression, are rate-limiting only after iron repletion has been achieved; this finding can have direct implications for the sequencing of anemia workup and treatment in clinical practice [2, 4, 5]. The study has several notable strengths: this is to our knowledge the first study to investigate the relationship between serum endostatin and anemia in older adults; it draws on a large multicenter European cohort of older adults ≥ 75 years, a population underrepresented in CKD and anemia research, thus enhancing the clinical relevance of study findings; furthermore, the prospective study design with a structured follow-up of two years allowed the investigation of both prevalent and incident anemia, providing preliminary evidence for a temporal relationship between endostatin and decline in hemoblobin concentration. However, it is important to recognize some limitations. First, the observational design precludes any conclusions regarding causality. Second, endostatin was measured at a single time point, which limits our ability to assess temporal trends or determine whether changes in endostatin may precede hemoglobin decline. Third, inflammatory biomarkers such as interleukin-6, hepcidin, and C-reactive protein, which are known to affect erythropoiesis and ESA responsiveness, were not included in the analysis, precluding the assessment of whether endostatin acts independently of systemic inflammation [49, 50]. Fourth, as participants with missing follow-up data differed systematically from those included in the longitudinal analysis, the longitudinal sample may underrepresent the most clinically complex individuals, leading to underestimation of the strength of the association between endostatin and anemia. Fifth, kidney function was estimated using the BIS equation, which relies on serum creatinine and may then lead to overestimation of GFR and underclassification of CKD severity in patients with decreased muscle mass. Finally, selection bias and limited generalizability must be considered: the study population consisted of older individuals who were able and willing to attend outpatient services and participate in a longitudinal research protocol, which may have led to the underrepresentation of frail, institutionalized, or acutely ill individuals. As a result, the findings may not be fully generalizable to all older adults with CKD or anemia. Conclusion In conclusion, in this large multicenter cohort of older adults, elevated circulating endostatin levels were independently associated with both lower hemoglobin levels and a higher prevalence and 2-year incidence of anemia. These associations were evident in individuals with chronic kidney disease, in those with preserved iron stores, and were stronger in those with ACR ≥ 30 mg/g and advanced age, suggesting that microvascular dysfunction and endothelial impairment may play a critical and previously underappreciated role in anemia pathogenesis in late life. Further research is needed to validate these findings in diverse and more vulnerable populations, clarify the causal pathways linking endostatin to anemia, and explore whether targeting vascular health could improve anemia-related outcomes in older adults with or without CKD. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1