Abstract
Radial-artery to cephalic-vein end-to-side arteriovenous fistulae (RCAVF) are the first-line vascular access for hemodialysis, yet early stenosis or thrombosis reduces its long-term patency. Indoxyl sulfate (IS), a protein-bound uremic toxin with diverse vascular effects, may impede fistula maturation. We conducted a single-center prospective cohort study enrolling 113 stage 5 chronic kidney disease patients who underwent RCAVF creation between February 2023 and August 2024. Patients were stratified by IS tertiles to assess fistula outcomes. Multivariable Cox regression, Kaplan-Meier analyses, and nomogram construction were used to identify risk factors and patency. Additionally, multiplex tyramide signal amplification immunofluorescence was applied to 40 samples of cephalic vein to quantify aryl hydrocarbon receptor (AHR) and tissue factor (TF) expression. After a median follow-up of 12 months, serum IS was higher in the dysfunction group (p = 0.005). Elevated IS (p = 0.013) and diabetes mellitus (p = 0.006) were independent risk factors for RCAVF dysfunction. Restricted cubic spline analysis showed a linear relationship between IS and risk. Kaplan-Meier curves revealed decreasing primary patency with increasing IS tertiles (p < 0.05), consistent in diabetic patients. The 1-year patency nomogram had good predictive performance (AUC = 0.87). Histopathology showed upregulated AHR and TF expression in veins from dysfunctional fistulas, with correlated fluorescence intensities (r = 0.60, p < 0.05), though neither correlated directly with serum IS (p > 0.05). In conclusion, elevated serum IS independently predicts RCAVF dysfunction, and the aberrant IS-AHR-TF signaling axis may contribute to its pathogenesis.
Keywords: Arteriovenous fistulas, aryl hydrocarbon receptor, indoxyl sulfate, tissue factor, vascular access dysfunction
1. Introduction
Hemodialysis, the primary renal replacement therapy for end-stage kidney disease (ESKD), relies on durable vascular access [1]. Current options include the radial-cephalic arteriovenous fistula (RCAVF), arteriovenous graft (AVG) and central venous catheter (CVC) [2,3]. Despite the superior longevity of RCAVF, 20–25% of AVFs/AVGs fail within one year because of venous stenosis (>50% lumen loss with flow < 500 mL min−1) and secondary thrombosis [4]. Access dysfunction accounts for roughly one-quarter of dialysis-related admissions, with a national rate of 1.78 hospitalizations per patient-year [5,6]. The underlying lesion is neointimal hyperplasia (NIH) of the outflow vein, driven by disturbed shear stress, endothelial injury and a pro-thrombotic milieu. Conventional clinical factors poorly predict early failure, highlighting the need for easily measurable biomarkers [7–10]. Recent studies have explored potential biomarkers for dialysis access dysfunction, among which fetuin-A could be used as a circulating biomarker to identify HD patients at greater risk for AV access dysfunction [11].
Indoxyl sulfate (IS), a protein-bound uremic toxin derived from tryptophan metabolism, is >90% albumin-bound and poorly removed by conventional dialysis [12,13]. Experimental studies show that IS activates the aryl-hydrocarbon receptor (AHR), up-regulates tissue factor (TF), induces reactive oxygen species, impairs endothelial repair and stimulates vascular smooth-muscle proliferation—processes that accelerate NIH and trigger thrombosis [14–21]. Observational data link higher IS concentrations to failure of AVF and AVG, yet prospective evidence for the prototypical RCAVF and in-situ confirmation of the IS–AHR–TF axis remain lacking [19–22].
Accordingly, it’s reasonable for us to hypothesize that serum IS levels can predict RCAVF dysfunction. In response, we conducted a single-center prospective cohort study to measure preoperative serum IS and monitor RCAVF stenosis or thrombosis within one year, develop and validate a risk prediction model, and quantify AHR and TF expression in excised cephalic vein tissue to clarify the mechanistic role of the IS–AHR–TF axis, aiming to inform early targeted interventions.
2. Materials and methods
2.1. Patient and group
Between February 2023 and August 2024, 113 consecutive stage-5 chronic kidney disease (CKD-5) patients who underwent primary radio-cephalic arteriovenous fistula (RCAVF) creation at our center were prospectively enrolled. Patients were categorized into two groups: (i) Patency group, defined as fistulas that permitted two-needle dialysis with a pump flow greater than 200 mL/min for at least six consecutive sessions; and (ii) Dysfunction group, including those who failed to meet these criteria due to stenosis and/or thrombosis. Additionally, patients were stratified into low, middle, and high tertiles based on serum IS levels. Follow-up assessments were conducted every three months for up to 24 months. Only the first dysfunction event per patient was included in the analysis (See Figure 1).
Figure 1.
Consolidated standards of reporting trials diagram.
The study was approved by the Ethics Committee of 900th Hospital of PLA Joint Logistic Support Force (IRB approval number 2022-053). All study procedures followed the Declaration of Helsinki.
2.2. Inclusion criteria
CKD stage 5;
age ≥ 18 years;
first autogenous vascular access created at our institution, located at the distal forearm radial–cephalic site;
spontaneous RCAVF maturation without adjunctive procedures;
no history of renal transplantation.
2.3. Exclusion criteria
prosthetic arteriovenous graft;
active malignancy or infection, shock, or haemodynamic instability;
skin ulcer/infection or haematoma/aneurysm at the fistula site;
revision surgery of a previous fistula;
active autoimmune disease or psychiatric disorder compromising compliance.
2.4. Clinical end-point
RCAVF dysfunction, defined as post-operative stenosis and/or thrombosis. Stenosis was characterized by clinical signs such as elevated venous pressure, difficult cannulation, inadequate dialysis, or abnormal physical findings. Ultrasound criteria included access flow < 400 mL/min, venous-to-arterial pressure ratio > 0.55, peak systolic velocity ≥ 500 cm/s, or ≥ 50% diameter reduction compared to adjacent segments. Diagnosis was supported by CDU, CTA, and digital subtraction angiography (DSA), the gold standard [23]. Thrombosis was identified by loss of thrill or bruit, vein hardening, absence of color-Doppler flow, and CTA/DSA evidence of no passage or stagnant contrast.
3. Study procedures
3.1. Baseline variables and laboratory tests
Demographic characteristics (sex, age, BMI, hypertension, diabetes, cardiovascular disease, erythropoietin use) were recorded. Laboratory parameters—haemoglobin, platelet count, serum creatinine, calcium, phosphate, calcium–phosphate product, intact parathyroid hormone, triglyceride, total cholesterol, Low Density Lipoprotein (LDL-C), High density lipoprotein cholesterol (HDL-C), albumin, C-reactive protein (CRP) and D-dimer—were measured in the hospital laboratory. estimated glomerular filtration rate (eGFR) was calculated using the CKD-EPI equation.
3.2. Serum sampling and indoxyl-sulfate assay
After 8–10 h overnight fasting, antecubital venous blood was drawn into plain tubes on the morning before surgery, clotted for 30 min at room temperature and centrifuged at 1,000 rpm for 20 min. Serum aliquots were stored at −20 °C. IS was quantified in batch using a commercial enzyme linked immunosorbent assay (ELISA) kit (Maidian Biotech, Fuzhou, China) according to the manufacturer’s protocol.
3.3. Histology of the cephalic vein
Pre-operative ultrasound confirmed vessel suitability. In 40 patients, a 1–5 mm remnant of the cephalic vein was excised at the anastomotic site during RCAVF creation.
Tyramide-signal-amplification (TSA) multiplex immunofluorescence: sections underwent EDTA antigen retrieval (pH 9.0) and sequential incubation with rabbit anti-AHR (1:2000) (AFW1451, Afanti, China) and anti-TF antibodies (AFW1378, Afanti, China), followed by HRP-polymer secondary antibody and TSA-520/570 fluorochromes; nuclei were counterstained with DAPI.
Images were captured on an optical or confocal microscope. Average fluorescence intensity (mean gray value) of AHR and TF within the intima-media region was calculated.
3.4. Statistical analysis
Analyses were performed with IBM SPSS Statistics 26.0. Normally distributed variables are expressed as mean ± SD and compared using the Student’s t-test or one-way ANOVA; skewed data are presented as median (IQR) and compared using the Mann–Whitney U or Kruskal–Wallis H test. Categorical variables were compared by χ2 or Fisher’s exact test. Kaplan–Meier curves with the log-rank test compared primary patency across IS tertiles. Independent predictors of RCAVF dysfunction were identified using Cox proportional-hazards modeling (covariates entered if p < 0.20 in univariable Cox); restricted cubic splines examined the dose–response relationship between IS and risk. A nomogram based on the Cox model was constructed, and its discrimination, calibration and clinical utility were evaluated with the area under the ROC curve (AUC), calibration plots and decision-curve analysis. A two-sided p < 0.05 indicated statistical significance.
4. Results
4.1. Clinical data comparison
A total of 113 ESKD patients underwent their first radio-cephalic arteriovenous fistula (RCAVF) creation at our center, including 83 men (73.45%) and 30 women (26.55%). Age ranged from 20 to 81 years (mean 54.09 ± 12.55 years). Patients were followed for 3–21 months (median 12 months). During follow-up, 23 patients (20.35%) experienced at least one RCAVF dysfunction event—nine cases of post-operative stenosis and 14 of thrombosis—all managed surgically (Table 1).
Table 1.
Baseline characteristics of ESKD patients stratified by serum indoxyl sulfate tertiles.
| Total (n = 113) | IS/low (n = 37) | IS/middle (n = 39) | IS/high (n = 37) | Statistic | p | |
|---|---|---|---|---|---|---|
| Sex (%) | χ² = 2.67 | 0.263 | ||||
| Female | 30 (26.55) | 8 (21.62) | 14 (35.90) | 8 (21.62) | ||
| Male | 83 (73.45) | 29 (78.38) | 25 (64.10) | 29 (78.38) | ||
| Age (years) | 54.09 ± 12.55 | 51.05 ± 14.21 | 55.10 ± 11.39 | 56.05 ± 11.70 | F = 1.68 | 0.191 |
| Hypertension (%) | χ² = 3.93 | 0.140 | ||||
| No | 18 (15.93) | 6 (16.22) | 3 (7.69) | 9 (24.32) | ||
| Yes | 95 (84.07) | 31 (83.78) | 36 (92.31) | 28 (75.68) | ||
| Diabetes Mellitus (%) | χ² = 2.59 | 0.274 | ||||
| No | 67 (59.29) | 24 (64.86) | 25 (64.10) | 18 (48.65) | ||
| Yes | 46 (40.71) | 13 (35.14) | 14 (35.90) | 19 (51.35) | ||
| Smoking (%) | χ² = 1.15 | 0.562 | ||||
| No | 92 (81.42) | 30 (81.08) | 30 (76.92) | 32 (86.49) | ||
| Yes | 21 (18.58) | 7 (18.92) | 9 (23.08) | 5 (13.51) | ||
| Antiplatelet drug (%) | – | 0.010 | ||||
| No | 99 (87.61) | 27 (72.97) | 37 (94.87) | 35 (94.59) | ||
| Yes | 14 (12.39) | 10 (27.03) | 2 (5.13) | 2 (5.41) | ||
| EPO (%) | χ² = 2.58 | 0.275 | ||||
| No | 93 (82.30) | 32 (86.49) | 29 (74.36) | 32 (86.49) | ||
| Yes | 20 (17.70) | 5 (13.51) | 10 (25.64) | 5 (13.51) | ||
| HIF-PHI (%) | χ² = 2.75 | 0.253 | ||||
| No | 55 (48.67) | 22 (59.46) | 16 (41.03) | 17 (45.95) | ||
| Yes | 58 (51.33) | 15 (40.54) | 23 (58.97) | 20 (54.05) | ||
| Statins (%) | χ² = 3.62 | 0.163 | ||||
| No | 79 (69.91) | 22 (59.46) | 31 (79.49) | 26 (70.27) | ||
| Yes | 34 (30.09) | 15 (40.54) | 8 (20.51) | 11 (29.73) | ||
| CVD (%) | χ² = 5.35 | 0.069 | ||||
| No | 91 (80.53) | 27 (72.97) | 36 (92.31) | 28 (75.68) | ||
| Yes | 22 (19.47) | 10 (27.03) | 3 (7.69) | 9 (24.32) | ||
| Dialysis (%) | – | 0.019 | ||||
| No | 12 (10.62) | 5 (13.51) | 7 (17.95) | 0 (0.00) | ||
| Yes | 101 (89.38) | 32 (86.49) | 32 (82.05) | 37 (100.00) | ||
| BMI (kg/m²) | 23.05 ± 3.33 | 22.78 ± 2.89 | 23.33 ± 3.49 | 23.04 ± 3.61 | F = 0.25 | 0.778 |
| SBP (mmHg) | 144.29 ± 19.63 | 140.62 ± 18.59 | 145.56 ± 22.33 | 146.62 ± 17.48 | F = 0.99 | 0.375 |
| DBP (mmHg) | 87.70 ± 13.22 | 87.14 ± 14.96 | 88.33 ± 11.96 | 87.59 ± 12.98 | F = 0.08 | 0.925 |
| Creatinine (μmol/L) | 795.94 ± 309.42 | 832.16 ± 349.37 | 736.54 ± 282.01 | 822.35 ± 293.20 | F = 1.11 | 0.334 |
| Cholesterol (mmol/L) | 4.35 ± 1.46 | 4.49 ± 2.13 | 4.42 ± 1.06 | 4.13 ± 0.91 | F = 0.63 | 0.535 |
| Triglyceride (mmol/L) | 1.65 ± 1.18 | 1.71 ± 1.07 | 1.59 ± 1.01 | 1.65 ± 1.45 | F = 0.10 | 0.904 |
| LDL-C (mmol/L) | 2.40 ± 0.94 | 2.55 ± 1.35 | 2.41 ± 0.71 | 2.22 ± 0.58 | F = 1.11 | 0.332 |
| HDL-C (mmol/L) | 0.98 ± 0.32 | 0.95 ± 0.26 | 1.05 ± 0.42 | 0.94 ± 0.24 | F = 1.58 | 0.211 |
| Potassium (mmol/L) | 4.24 ± 0.73 | 4.35 ± 0.74 | 4.15 ± 0.59 | 4.23 ± 0.86 | F = 0.72 | 0.487 |
| Calcium (mmol/L) | 2.10 ± 0.28 | 2.09 ± 0.19 | 2.05 ± 0.26 | 2.18 ± 0.36 | F = 2.07 | 0.131 |
| Phosphorus (mmol/L) | 1.65 ± 0.47 | 1.76 ± 0.32 | 1.61 ± 0.61 | 1.57 ± 0.40 | F = 1.80 | 0.171 |
| Calcium-phosphorus product | 42.65 ± 11.38 | 45.60 ± 9.22 | 39.98 ± 10.66 | 42.51 ± 13.47 | F = 2.37 | 0.098 |
| iPTH (pg/ml) | 380.94 ± 243.45 | 368.72 ± 217.76 | 392.53 ± 253.39 | 380.95 ± 262.49 | F = 0.09 | 0.915 |
| Albumin (g/L) | 34.13 ± 5.19 | 34.00 ± 5.05 | 34.80 ± 5.80 | 33.56 ± 4.70 | F = 0.56 | 0.573 |
| Hemoglobin (g/L) | 86.97 ± 16.05 | 88.00 ± 15.82 | 84.28 ± 15.71 | 88.78 ± 16.69 | F = 0.86 | 0.427 |
| platelet count (*109/L) | 199.50 ± 66.07 | 206.62 ± 67.78 | 199.97 ± 74.61 | 191.89 ± 54.76 | F = 0.46 | 0.634 |
| INR | 1.01 ± 0.08 | 1.02 ± 0.07 | 1.00 ± 0.08 | 1.02 ± 0.08 | F = 0.70 | 0.501 |
| Fibrinogen (g/L) | 4.22 ± 1.19 | 4.36 ± 1.37 | 4.24 ± 1.17 | 4.06 ± 1.00 | F = 0.56 | 0.571 |
| D-dimer (mg/L) | 1.95 ± 2.62 | 1.65 ± 2.18 | 2.49 ± 3.25 | 1.69 ± 2.23 | F = 1.24 | 0.293 |
| CRP (mg/L) | 9.60 ± 6.99 | 10.57 ± 9.80 | 8.83 ± 5.03 | 9.45 ± 5.27 | F = 0.60 | 0.551 |
| eGFR (mL/min/1.73 m2) | 9.00 ± 2.67 | 8.47 ± 2.53 | 9.71 ± 2.81 | 8.77 ± 2.56 | F = 2.30 | 0.105 |
| Myoglobin (μg/L) | 351.76 ± 233.41 | 318.19 ± 189.72 | 313.04 ± 222.10 | 426.13 ± 270.13 | F = 2.89 | 0.060 |
| Ultrasensitive troponin (μg/L) | 0.07 ± 0.27 | 0.03 ± 0.04 | 0.05 ± 0.12 | 0.12 ± 0.46 | F = 1.00 | 0.370 |
| Natriuretic peptide (pg/mL) | 340.22 ± 660.33 | 181.29 ± 187.41 | 326.13 ± 456.12 | 514.01 ± 1022.07 | F = 2.42 | 0.093 |
| Cardiac ejection fraction (%) | 63.34 ± 5.23 | 63.88 ± 4.56 | 62.43 ± 6.13 | 63.75 ± 4.82 | F = 0.90 | 0.411 |
| IS (ng/mL) | 19.15 ± 11.09 | 10.10 ± 2.85 | 15.51 ± 1.55 | 32.04 ± 10.12 | F = 130.85 | <.001*** |
Note: * indicates p < 0.05, ** indicates p < 0.01, ***, p < 0.001; CRP: C-reactive protein; BMI: body mass index; SBP: systolic blood pressure; DBP: diastolic blood pressure; LDL-C: low-density lipoprotein; HDL-C: high-density lipoprotein; iPTH: whole segment parathyroid hormone; eGFR: estimated glomerular filtration rate; EPO erythropoietin; HIF-PHI: hypoxia-inducible factor prolyl hydroxylase inhibitor; CVD: cardiovascular disease (coronary artery disease, peripheral vascular, and cerebrovascular disease); INR: International Normalized Ratio; IS: indole sulfate.
IS was significantly higher in the dysfunction group than in the patency group (26.53 ± 13.70 vs 17.26 ± 9.52 ng mL−1, p = 0.005). There were no statistical differences among the groups in demographics, laboratory data, etc. Median IS levels remained higher in dysfunction cases overall (Figure 2A) and within the diabetic subgroup (Figure 2B).
Figure 2.
(A) Comparison of serum IS levels between dysfunction (+) and patency (–) groups.(B) Stratified analysis by diabetes status: A = non-diabetic patients; B = diabetic patients; + = dysfunction group; – = patency group.
4.2. Follow-up analysis
4.2.1. Cox regression for independent predictors of RCAVF dysfunction
Univariable Cox proportional-hazards analysis revealed that both a history of diabetes mellitus and IS were significantly associated with RCAVF dysfunction (p < 0.05; Table 2). Variables with p < 0.20 in the univariable screen—including HDL-cholesterol (HDL-C), serum calcium, CRP and brain natriuretic peptide (BNP)—were subsequently entered into a multivariable model. In this model, diabetes mellitus (p = 0.006) and serum IS (p = 0.013) emerged as independent risk factors for RCAVF dysfunction (Table 2).
Table 2.
Univariate and multivariate cox regression analysis of independent risk factors for RCAVF dysfunction.
| Variable | One-way Cox regression analysis |
Multifactor Cox regression analysis |
||||
|---|---|---|---|---|---|---|
| HR | 95%CI | p | HR | 95%CI | p | |
| Sex | ||||||
| Female | 1.00 | Reference | ||||
| Male | 0.91 | 0.36–2.31 | 0.837 | |||
| Age | 0.98 | 0.95–1.02 | 0.316 | |||
| Hypertension | ||||||
| No | 1.00 | Reference | ||||
| Yes | 0.60 | 0.22–1.63 | 0.320 | |||
| Diabetes Mellitus | ||||||
| No | 1.00 | Reference | 1.00 | Reference | ||
| Yes | 4.95 | 1.94–12.63 | <.001*** | 4.12 | 1.50–11.34 | 0.006** |
| Smoking | ||||||
| No | 1.00 | Reference | ||||
| Yes | 0.54 | 0.16–1.83 | 0.322 | |||
| Antiplatelet drug | ||||||
| No | 1.00 | Reference | ||||
| Yes | 0.30 | 0.04—2.24 | 0.242 | |||
| EPO | ||||||
| No | 1.00 | Reference | ||||
| Yes | 0.42 | 0.10–1.78 | 0.238 | |||
| HIF-PHI | ||||||
| No | 1.00 | Reference | ||||
| Yes | 0.87 | 0.38–1.97 | 0.741 | |||
| Statins | ||||||
| No | 1.00 | Reference | ||||
| Yes | 0.66 | 0.25–1.79 | 0.418 | |||
| CVD | ||||||
| No | 1.00 | Reference | ||||
| Yes | 0.63 | 0.19–2.14 | 0.463 | |||
| Dialysis (%) | ||||||
| No | 1.00 | Reference | ||||
| Yes | 0.97 | 0.22–4.18 | 0.964 | |||
| BMI | 0.98 | 0.87–1.11 | 0.778 | |||
| SBP | 1.00 | 0.98–1.02 | 0.961 | |||
| DBP | 0.99 | 0.96–1.02 | 0.542 | |||
| Creatinine | 1.00 | 1.00–1.00 | 0.689 | |||
| Cholesterol | 1.07 | 0.80–1.41 | 0.658 | |||
| Triglyceride | 0.92 | 0.60–1.40 | 0.685 | |||
| LDL-C | 1.16 | 0.77–1.74 | 0.473 | |||
| HDL-C | 0.25 | 0.05–1.30 | 0.098 | |||
| Potassium | 1.17 | 0.67–2.06 | 0.578 | |||
| Calcium | 2.27 | 0.76–6.83 | 0.143 | |||
| Phosphorus | 0.58 | 0.20–1.70 | 0.322 | |||
| Calcium-phosphorus product | 1.00 | 0.96–1.03 | 0.901 | |||
| iPTH | 1.00 | 1.00–1.00 | 0.232 | |||
| Albumin | 0.98 | 0.90–1.06 | 0.635 | |||
| Hemoglobin | 1.01 | 0.98–1.03 | 0.582 | |||
| Platelet count | 1.00 | 0.99–1.01 | 0.932 | |||
| INR | 0.40 | 0.00–76.57 | 0.732 | |||
| Fibrinogen | 0.79 | 0.54–1.16 | 0.226 | |||
| D-dimer | 1.03 | 0.89–1.18 | 0.706 | |||
| CRP | 1.03 | 0.99–1.08 | 0.153 | |||
| eGFR | 1.08 | 0.93–1.26 | 0.311 | |||
| Myoglobin | 1.00 | 1.00–1.00 | 0.863 | |||
| Ultrasensitive troponin (μg/L) | 0.40 | 0.01–25.55 | 0.667 | |||
| Natriuretic peptide | 1.00 | 1.00–1.00 | 0.184 | |||
| Cardiac ejection fraction | 0.96 | 0.90–1.04 | 0.329 | |||
| IS | 1.05 | 1.02–1.08 | <.001*** | 1.04 | 1.01–1.07 | 0.013* |
Note: * indicates p < 0.05, ** indicates p < 0.01, ***, p < 0.001; CRP: C-reactive protein; BMI: body mass index; SBP: systolic blood pressure; DBP: diastolic blood pressure; LDL-C: low-density lipoprotein; HDL-C: high-density lipoprotein; iPTH: whole segment parathyroid hormone; eGFR: estimated glomerular filtration rate; EPO erythropoietin; HIF-PHI: hypoxia-inducible factor prolyl hydroxylase inhibitor; CVD: cardiovascular disease (coronary artery disease, peripheral vascular, and cerebrovascular disease); INR: International Normalized Ratio; IS: indole sulfate.
4.2.1.1. Adjusted models
Multivariable Cox regression was performed using three incremental models (Models 1–3; Table S1). After simultaneous adjustment for diabetes mellitus, HDL-cholesterol, serum calcium, CRP and BNP, IS remained an independent predictor of RCAVF dysfunction (HR = 1.04, 95% CI 1.01–1.07, p = 0.013).
4.3. Restricted cubic-spline analysis
Restricted cubic-spline modeling demonstrated a progressive increase in the predicted probability of RCAVF dysfunction with rising serum IS levels. The overall test for association was significant (P_overall = 0.037), whereas the test for non-linearity was not (P_non-linear = 0.264), indicating a primarily linear relationship (Figure 3). In summary, higher serum IS may linearly elevate the risk of RCAVF dysfunction.
Figure 3.
Linear association between serum IS levels and RCAVF dysfunction.
The red smoothed curve represents the predicted probability trend of RCAVF dysfunction based on the model, and the shaded area indicates the 95% confidence interval (95% CI). The model was adjusted for potential confounders including diabetes mellitus, high-density lipoprotein (HDL), serum calcium, C-reactive protein (CRP), and B-type natriuretic peptide (BNP).
4.4. Kaplan–meier analysis of RCAVF patency across serum is tertiles
Kaplan–Meier analysis of RCAVF patency across serum IS tertiles Patients were divided into three groups according to tertiles of baseline serum IS concentrations (overall range 3.56–55.41 ng/ml): low tertile < 12.46 ng/ml, middle tertile 12.46–24.69 ng/ml, and high tertile > 24.69 ng/ml. Kaplan–Meier analysis showed a step-wise decline in primary RCAVF patency across low, middle, and high IS tertiles, with the steepest drop in the high-IS group (log-rank p < 0.001; Figure 4A). This gradient persisted among diabetic patients (log-rank p < 0.001; Figure 4C) but disappeared in non-diabetic patients (log-rank p = 0.531; Figure 4B).
Figure 4.
Kaplan-Meier curves of RCAVF patency stratified by serum IS tertiles.
(A) All patients stratified by serum indoxyl sulfate (IS) tertiles. (B) Non-diabetic patients stratified by serum IS tertiles. (C) Diabetic patients stratified by serum IS tertiles. p-values were calculated using the log-rank test. Groups: 1 = Low IS level; 2 = Medium IS level; 3 = High IS level.
4.5. Development and evaluation of a nomogram for RCAVF patency
Using variables retained in the multivariable Cox model, a nomogram was constructed to estimate one-year RCAVF patency in ESKD patients (Figure 5). The model demonstrated excellent discrimination with an AUC of 0.87 (95% CI 0.78–0.97) (Figure 6A), showed close agreement between predicted and observed probabilities on the calibration plot (Figure 6B), and provided greater net benefit than treat-all or treat-none strategies across threshold probabilities of 10–60% in decision-curve analysis (Figure 6C), indicating substantial clinical utility for early risk stratification and tailored follow-up or intervention.
Figure 5.
Nomogram model for predicting 1-year RCAVF patency in ESKD patients.
In the nomogram, first locate a patient’s value for each predictor on the corresponding axis: presence of diabetes and serum IS concentration (ng/mL). Draw a vertical line from each value up to the top “Points” scale to determine the number of points contributed by that predictor. Sum the points for all predictors to obtain the value on the “Total Points” axis, then draw a vertical line downward from this total to the “1-year patency rate” axis to estimate the individual probability of 1-year primary patency.
Figure 6.
Evaluation of the predictive performance and clinical utility of the nomogram model. (A) Receiver operating characteristic (ROC) curve for the nomogram model predicting 1-year RCAVF patency in ESKD patients. The area under the curve (AUC) was 0.87 (95% CI: 0.78–0.97). (B) Calibration curve of the nomogram model for predicting 1-year RCAVF patency. The x-axis represents the predicted probability of patency, and the y-axis represents the observed probability. (C) Decision curve analysis (DCA) of the nomogram model. The x-axis indicates the threshold probability at which a patient would be classified as high risk according to the predicted probability of 1-year fistula dysfunction. The y-axis shows the net benefit, calculated by weighting true positives against false positives at each threshold. The blue line represents the net benefit of the nomogram, the red line represents the strategy of treating all patients, and the green line represents the strategy of treating no patients. The nomogram is considered clinically useful over the range of threshold probabilities where its curve lies above both the treat-all and treat-none lines.
5. Correlation between AHR/TF expression and RCAVF dysfunction
Tyramide-signal-amplification multiplex immunofluorescence demonstrated that the mean fluorescence intensities of both AHR and TF within the intima–media region of the cephalic vein were significantly higher in the dysfunction group than in the patency group (each p < 0.0001; Figure 8A and B), with the dysfunction specimens exhibiting broader and more intense positive signals for both proteins (Figure 7).
Figure 8.
Comparison of AHR and TF fluorescence intensity in cephalic vein tissues between RCAVF dysfunction and patency groups. ****p < 0.0001; + = dysfunction group; – = patency group.
Figure 7.
Representative confocal microscopy images of immunofluorescence staining in outflow vein tissues from the dysfunction and patency groups.
The first column shows DAPI staining intensity, indicating nuclei (blue). The second column illustrates AHR staining intensity, with AHR-positive cells shown in green. The third column depicts TF staining intensity, with TF-positive cells shown in red. The fourth column shows merged images. These images demonstrate differential expression levels of AHR and TF proteins in the outflow veins between the two groups.
Linear regression indicated a significant positive correlation between AHR and TF fluorescence intensities (r = 0.60, p < 0.001), whereas neither protein showed a significant relationship with serum IS levels (p > 0.05; Table S2). A correlation heatmap constructed from 14 variables—AHR, TF, serum IS, diabetes status, nutritional markers (haemoglobin, albumin), coagulation markers (platelet count, fibrinogen), calcium–phosphate metabolism (parathyroid hormone, calcium, phosphate), lipid profile (triglyceride, cholesterol) and inflammation (C-reactive protein)—revealed that aside from the AHR–TF pair the coefficients largely ranged from −0.28 to 0.26, indicating generally weak inter-variable correlations and suggesting relative independence of these risk factors within the study population (Figures 8, 9).
Figure 9.
Correlation heatmap analysis among serum IS levels, fluorescence intensity of AHR/TF in the cephalic vein, and clinicopathological characteristics.
Colors range from dark purple (–1) to bright yellow (+1), with values closer to +1 indicating stronger positive correlations, values closer to –1 indicating stronger negative correlations, and values near 0 indicating little or no linear correlation.
6. Discussion
Hemodialysis remains the mainstay therapy for chronic kidney failure [1]. Among the available vascular accesses, an autogenous arteriovenous fistula (AVF) is preferred for maintenance hemodialysis (MHD) patients [2,3]; yet stenosis and thrombosis frequently degrade its performance, compromising dialysis adequacy and quality of life [24]. The present prospective study was designed to clarify the role of serum IS in radio-cephalic AVF (RCAVF) dysfunction—including post-operative stenosis and thrombosis—to develop a risk-prediction model, and to explore histological as well as molecular correlates in the cephalic vein. We had hypothesized that serum IS levels can predict RCAVF dysfunction.
Using multivariable Cox regression, we identified pre-operative serum IS and diabetes mellitus as independent predictors of RCAVF dysfunction (IS: HR = 1.04 per ng mL, 95% CI 1.02–1.07, p = 0.002; diabetes: HR = 4.60, 95% CI 1.78–11.84, p = 0.002). A nomogram incorporating only these two variables predicted one-year RCAVF patency with excellent discrimination (AUC = 0.87), good calibration, and substantial net benefit on decision-curve analysis. These metrics underscore its potential for bedside risk stratification in end-stage renal disease (ESKD) patients.
Kaplan-Meier curves revealed a marked inverse relationship between IS tertiles and fistula patency, with a graded loss over time (log-rank p < 0.001). Strikingly, this gradient was preserved in diabetic patients (p < 0.001) but vanished among non-diabetics (p = 0.531), suggesting that diabetes amplifies IS-related vascular toxicity. Restricted cubic-spline modeling confirmed an essentially linear dose–response (overall p = 0.037; Pnon-linear = 0.264). We found that a diabetes-dependent effect of IS on AVF patency, highlighting a disease-toxin synergy that may explain the disproportionately high AVF failure rates in diabetic ESKD populations. Diabetes as an amplifier A meta-analysis reported higher AVF failure in diabetics (OR = 1.68, 95% CI 1.43–1.98) [25], congruent with our finding that diabetes raises dysfunction risk four-fold. Hyperglycemia, advanced glycation end-products, platelet hyper-reactivity, endothelial damage, and a predisposition to vascular calcification converge to accelerate intimal–medial thickening and thrombosis in diabetic CKD.
Hypertension and sex differences subgroup analysis suggested a higher hazard ratio for dysfunction in hypertensive versus normotensive patients (HR = 1.17 vs 1.04) and in women versus men (HR = 1.09 vs 1.04), although interaction terms were non-significant, possibly because of limited power. Prior literature on sex differences is conflicting: some studies report poorer AVF maturation in women despite similar intimal fibrosis, whereas others document higher one-year patency in women [26,27]. Experimental data indicate sex-dependent AHR activation [27]; thus, IS-mediated AHR signaling might partially underlie these discrepancies.
Endothelial cells normally suppress inflammation and thrombosis [28], but uremic toxins like IS and indole-3-acetic acid (IAA) activate AHR, upregulate TF, and induce pro-coagulant activity [18,19,29,30]. IS also promotes vascular smooth-muscle cell (VSMC) calcification via multiple pathways [31–33] and stimulates both VSMC migration (key in AVF neointimal hyperplasia [34] and proliferation. Combined with disturbed anastomotic haemodynamics, this creates a cycle of intimal thickening, thrombosis, and stenosis. The linear association between IS and AVF failure echoes in-vitro findings that IS dose-dependently induces endothelial injury, oxidative stress, and VSMC proliferation [35]. IS also activates AHR, up-regulating TF and fostering a pro-thrombotic state [18]. In our study, tyramide signal amplification multiplex immunofluorescence showed significantly higher AHR and TF intensities in the intima-media region of dysfunctional versus patent RCAVFs (both p < 0.0001). AHR intensity correlated positively with TF (r = 0.60, p < 0.05).
Although AHR and TF intensities did not correlate with circulating IS in our material, several factors may account for this discrepancy: ligand–receptor affinity saturation, synergistic activation by other uremic toxins (e.g. kynurenine, IAA), or local haemodynamic triggers that prime AHR signaling independently of systemic IS concentration. Meanwhile, recent work has shown that endothelial microparticles carrying TF are released after AHR activation by IS or IAA, enhancing coagulation potential [19]. Collectively, these data position the IS–AHR–TF axis as a plausible driver of thrombosis and intimal hyperplasia in RCAVF. Previous studiesl [20] have demonstrated that both cellular and animal experiments implicate the IS–AhR–TF pathway in endothelial thrombogenesis; however, robust clinical evidence remains lacking. Our findings suggest that the IS–AhR–TF pathway may contribute to the development of AVF dysfunction, including thrombosis and stenosis, and this study provides substantial supporting evidence. Admittedly, this study has several limitations. First, the sample size was relatively small, and the single-center design may limit the generalizability and representativeness of the findings, as patients with secondary AVFs or prosthetic grafts were not included. Second, the median follow-up duration was 12 months, which is relatively short and may not adequately capture the long-term impact of serum IS on RCAVF dysfunction. Third, the immunofluorescence technique used in this study is not a fully quantitative method, whereas Western blotting can more rigorously validate differential expression.
Supplementary Material
Acknowledgements
The authors thank the staffs from the 900th Hospital of Joint Logistic Support Force for their help. Fujian Provincial Department of Science and Technology. QS, PZ, QW contributed to the study conception and design. Data collection and organization were conducted by PZ, QW, YL, MQ. The statistical analyses were performed by QS, XW. The first draft of the manuscript was written by QS. All authors read and approved the final manuscript. All authors searched the literature.
Correction Statement
This article has been corrected with minor changes. These changes do not impact the academic content of the article.
Funding Statement
This study was supported by Guiding Project of Fujian Provincial Science and Technology Department (No.2025Y0052) and the Fujian Clinical Medical Research Center for Immune Kidney Disease (NO.2021Y2016) and Joint logistic support force joint logistics medical high-quality specialty (LQYZ-SZ).
Ethics approval
The study protocol received approval from the Ethics Committee of 900th Hospital of PLA Joint Logistic Support Force (IRB approval number 2022-053). Written informed consent was obtained from all participants.
Disclosure of interest
No potential competing interest was reported by the author(s).
Data availability statement
The datasets used and analyzed during the current study available from the corresponding author on reasonable request.
References
- 1.Basile C, Davenport A, Mitra S, et al. Frontiers in hemodialysis: innovations and technological advances. Artif Organs. 2021;45(2):175–182. doi: 10.1111/aor.13798. [DOI] [PubMed] [Google Scholar]
- 2.Wei S, Jiao J, Yu Y, et al. Long-term arteriovenous fistula prognosis for maintenance hemodialysis patients who accepted PIRRT by using arteriovenous fistula. Int J Artif Organs. 2023;46(4):195–201. doi: 10.1177/03913988231162384. [DOI] [PubMed] [Google Scholar]
- 3.Li Y, Cui W, Wang J, et al. Comparative efficacy of five balloons for treating autogenous arteriovenous fistula stenosis: a Bayesian network meta-analysis. Ann Palliat Med. 2022;11(8):2574–2585. doi: 10.21037/apm-21-2898. [DOI] [PubMed] [Google Scholar]
- 4.Huber TS, Berceli SA, Scali ST, et al. Arteriovenous Fistula Maturation, Functional Patency, and Intervention Rates. JAMA Surg. 2021;156(12):1111–1118. doi: 10.1001/jamasurg.2021.4527. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.MacAskill MG, Watson DG, Ewart MA, et al. Improving arteriovenous fistula patency: transdermal delivery of diclofenac reduces cannulation-dependent neointimal hyperplasia via AMPK activation[J]. Vascul Pharmacol. 2015;71:108–115. doi: 10.1016/j.vph.2015.02.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Bilgic MA, Yilmaz H, Bozkurt A, et al. Relationship of late arteriovenous fistula stenosis with soluble E-selectin and soluble EPCR in chronic hemodialysis patients with arteriovenous fistula. Clin Exp Nephrol. 2015;19(1):133–139. doi: 10.1007/s10157-014-0955-4. [DOI] [PubMed] [Google Scholar]
- 7.Roy-Chaudhury P, Sukhatme VP, Cheung AK.. Hemodialysis Vascular Access Dysfunction. Journal of the American Society of Nephrology. 2006;17(4):1112–1127. doi: 10.1681/ASN.2005050615. [DOI] [PubMed] [Google Scholar]
- 8.Schwab SJ, Oliver MJ, Suhocki P, et al. Hemodialysis arteriovenous access: detection of stenosis and response to treatment by vascular access blood flow[J]. Kidney Int. 2001;59(1):358–362. doi: 10.1046/j.1523-1755.2001.00498.x. [DOI] [PubMed] [Google Scholar]
- 9.Moist LM, Churchill DN, House AA, et al. Regular Monitoring of Access Flow Compared with Monitoring of Venous Pressure Fails to Improve Graft Survival. J Am Soc Nephrol. 2003;14(10):2645–2653. doi: 10.1097/01.asn.0000089562.98338.60. [DOI] [PubMed] [Google Scholar]
- 10.Neuen BL, Gunnarsson R, Webster AC, et al. Predictors of Patency after Balloon Angioplasty in Hemodialysis Fistulas: a Systematic Review. J Vasc Interv Radiol. 2014;25(6):917–924. doi: 10.1016/j.jvir.2014.02.010. [DOI] [PubMed] [Google Scholar]
- 11.Roca-Tey R, Ramírez de Arellano M, González-Oliva JC, et al. Is fetuin-A a biomarker of dialysis access dysfunction? [J Vasc Access. 2023;24(3):458–464. doi: 10.1177/11297298211035846. [DOI] [PubMed] [Google Scholar]
- 12.Liu W-C, Tomino Y, Lu K-C.. Impacts of Indoxyl Sulfate and p-Cresol Sulfate on Chronic Kidney Disease and Mitigating Effects of AST-120. Toxins (Basel). 2018;10(9):367. doi: 10.3390/toxins10090367. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Madero M, Cano KB, Campos I, et al. Removal of Protein-Bound Uremic Toxins during Hemodialysis Using a Binding Competitor. Clin J Am Soc Nephrol. 2019;14(3):394–402. doi: 10.2215/CJN.05240418. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Dou L, Jourde‐chiche N, Faure V, et al. The uremic solute indoxyl sulfate induces oxidative stress in endothelial cells[J]. J Thromb Haemost. 2007;5(6):1302–1308. doi: 10.1111/j.1538-7836.2007.02540.x. [DOI] [PubMed] [Google Scholar]
- 15.Dou L, Bertrand E, Cerini C, et al. The uremic solutes p-cresol and indoxyl sulfate inhibit endothelial proliferation and wound repair[J]. Kidney Int. 2004;65(2):442–451. doi: 10.1111/j.1523-1755.2004.00399.x. [DOI] [PubMed] [Google Scholar]
- 16.Yu M, Kim YJ, Kang D-H.. Indoxyl Sulfate–Induced Endothelial Dysfunction in Patients with Chronic Kidney Disease via an Induction of Oxidative Stress. Clin J Am Soc Nephrol. 2011;6(1):30–39. doi: 10.2215/CJN.05340610. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Yamamoto H, Tsuruoka S, Ioka T, et al. Indoxyl sulfate stimulates proliferation of rat vascular smooth muscle cells. Kidney Int. 2006;69(10):1780–1785. doi: 10.1038/sj.ki.5000340. [DOI] [PubMed] [Google Scholar]
- 18.Bouabdallah J, Zibara K, Issa H, et al. Endothelial cells exposed to phosphate and indoxyl sulphate promote vascular calcification through interleukin-8 secretion. Nephrol Dial Transplant. 2019;34(7):1125–1134. doi: 10.1093/ndt/gfy325. [DOI] [PubMed] [Google Scholar]
- 19.Gondouin B, Cerini C, Dou L, et al. Indolic uremic solutes increase tissue factor production in endothelial cells by the aryl hydrocarbon receptor pathway. Kidney Int. 2013;84(4):733–744. doi: 10.1038/ki.2013.133. [DOI] [PubMed] [Google Scholar]
- 20.Shivanna S, Kolandaivelu K, Shashar M, et al. The Aryl Hydrocarbon Receptor is a Critical Regulator of Tissue Factor Stability and an Antithrombotic Target in Uremia. J Am Soc Nephrol. 2016;27(1):189–201. doi: 10.1681/ASN.2014121241. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Shashar M, Belghasem ME, Matsuura S, et al. Targeting STUB1–tissue factor axis normalizes hyperthrombotic uremic phenotype without increasing bleeding risk[J]. Sci Transl Med. 2017;9(417):eaam8475. doi: 10.1126/scitranslmed.aam8475. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Kolachalama VB, Shashar M, Alousi F, et al. Uremic Solute-Aryl Hydrocarbon Receptor-Tissue Factor Axis Associates with Thrombosis after Vascular Injury in Humans. J Am Soc Nephrol. 2018;29(3):1063–1072. doi: 10.1681/ASN.2017080929. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Wo K, Morrison BJ, Harada RN.. Developing Duplex Ultrasound Criteria for Diagnosis of Arteriovenous Fistula Stenosis. J Vasc Surg. 2017;65(6):1863–1864. doi: 10.1016/j.jvs.2017.04.006. [DOI] [PubMed] [Google Scholar]
- 24.Meola M, Marciello A, DI Salle G, et al. Ultrasound evaluation of access complications: thrombosis, aneurysms, pseudoaneurysms and infections. J Vasc Access. 2021;22(1_suppl):71–83. doi: 10.1177/11297298211018062. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Yan Y, Ye D, Yang L, et al. A meta-analysis of the association between diabetic patients and AVF failure in dialysis. Ren Fail. 2018;40(1):379–383. doi: 10.1080/0886022X.2018.1456464. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.ANGELICI L, MARINO C, UMBRO I, et al. Gender Disparities in Vascular Access and One-Year Mortality among Incident Hemodialysis Patients: an Epidemiological Study in Lazio Region, Italy. J Clin Med 2021, 10(21, 5116). doi: 10.3390/jcm10215116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Palzkill VR, Tan J, Yang Q, et al. Deletion of the Aryl Hydrocarbon Receptor in Endothelial Cells Improves Ischemic Angiogenesis in Chronic Kidney Disease. Am J Physiol Heart Circ Physiol. 2024;326(1):H44–H60. H44-H60. doi: 10.1152/physiol.2024.39.S1.900. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Endemann DH, Schiffrin EL.. Endothelial Dysfunction. J Am Soc Nephrol. 2004;15(8):1983–1992. doi: 10.1097/01.ASN.0000132474.50966.DA. [DOI] [PubMed] [Google Scholar]
- 29.Addi T, Poitevin S, McKay N, et al. Mechanisms of tissue factor induction by the uremic toxin indole-3 acetic acid through aryl hydrocarbon receptor/nuclear factor-kappa B signaling pathway in human endothelial cells. Arch Toxicol. 2019;93(1):121–136. doi: 10.1007/s00204-018-2328-3. [DOI] [PubMed] [Google Scholar]
- 30.Chitalia VC, Shivanna S, Martorell J, et al. Uremic solute–aryl hydrocarbon receptor–tissue factor axis contributes to thrombosis after vascular injury. Circulation. 2013;127(3):365–376. doi: 10.1161/CIRCULATIONAHA.112.118174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Wu Y, Han X, Wang L, et al. Indoxyl sulfate promotes vascular smooth muscle cell calcification via the JNK/Pit-1 pathway. Ren Fail. 2016;38(10):1702–1710. doi: 10.3109/0886022X.2016.1155397. [DOI] [PubMed] [Google Scholar]
- 32.Muteliefu G, Enomoto A, Jiang P, et al. Indoxyl sulphate induces oxidative stress and the expression of osteoblast-specific proteins in vascular smooth muscle cells. Nephrol Dial Transplant. 2009;24(7):2051–2058. doi: 10.1093/ndt/gfn757. [DOI] [PubMed] [Google Scholar]
- 33.TORREMADé N, Bozic M, Panizo S, et al. Vascular Calcification Induced by Chronic Kidney Disease Is Mediated by an Increase of 1α-Hydroxylase Expression in Vascular Smooth Muscle Cells. J Bone Miner Res. 2016;31(10):1865–1876. doi: 10.1002/jbmr.2852. [DOI] [PubMed] [Google Scholar]
- 34.Kokubo T, Ishikawa N, Uchida H, et al. CKD accelerates development of neointimal hyperplasia in arteriovenous fistulas. J Am Soc Nephrol. 2009;20(6):1236–1245. doi: 10.1681/ASN.2007121312. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Hénaut L, Mary A, Chillon J-M, et al. The Impact of Uremic Toxins on Vascular Smooth Muscle Cell Function. Toxins (Basel). 2018;10(6):218. doi: 10.3390/toxins10060218. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets used and analyzed during the current study available from the corresponding author on reasonable request.