Depressive symptoms in chronic kidney disease: the hidden role of uremic toxins Authors: Hélène Levassort, Sophie Liabeuf, Julie Boucquemont, Gaye Hafez, Solene M Laville, Celine Lange, Luc Frimat, Christian Combe, Denis Fouque, Maurice Laville, Christian Jacquelinet, Yves-Edouard Herpe, Lucile Montalescot, Islam Amine Larabi, Natalia Alencar de Pinho, Ziad A Massy, Jean-Claude Alvarez, Marion Pépin, on behalf of CKD-REIN Published: 2026 ABSTRACT AbstractBackgroundDepression is closely linked to neuroinflammation, and in chronic kidney disease, the accumulation of uremic toxins (UTs) may promote neuroinflammatory processes through the activation of inflammatory pathways.ObjectiveThe objective of this longitudinal study was to evaluate the association between UTs and depressive symptoms in a large, well-characterised cohort of non-dialysed adults with chronic kidney disease, a population in which this relationship had not yet been investigated.MethodsThe Chronic Kidney Disease-Renal Epidemiology and Information Network (CKD-REIN) cohort comprised 3033 CKD stage 2–5 patients with 5 years of follow-up. Changes over time in depressive symptoms were assessed on the Center for Epidemiologic Studies Depression (CESD) scale. Mixed models were used to examine associations between the change in the CESD score and baseline levels of the UTs phenylacetylglutamine (PAG), indoxyl sulphate (IS), indole-3-acetic acid (IAA), p-cresyl sulphate, kynurenine, kynurenic acid and trimethylamine N-oxide.Findings2165 patients were included (median age: 68 years; mean estimated glomerular filtration rate (eGFR): 35 mL/min/1.73 m²; median baseline CESD score: 7; mean follow-up time: 4.0 years). After adjustment for confounders, the CESD score was found to have increased by 0.11 (0.04 to 0.18) points per year. A doubling in PAG level was associated with an additional increase of 0.06 (0.01 to 0.18) points per year in the CESD. A doubling in IS and IAA levels was associated with a higher mean CESD score (0.20 (0.01 to 0.39) for IS and 0.28 (0.05 to 0.50) for IAA) but not with the score’s change over time.ConclusionPAG was significantly associated with changes in CESD score over time. Higher IS and IAA levels were linked to higher mean scores. Further studies are needed to confirm these results and to determine whether lowering serum UT levels would help to manage depressive symptoms in patients with CKD.Clinical implicationsFurther research is needed to clarify the role of UTs; a better understanding could uncover novel metabolic pathways and inform new therapeutic strategies to complement the management of depressive symptoms.Keywords: Depression WHAT IS ALREADY KNOWN ON THIS TOPIC Depression is strongly linked to neuroinflammation. In chronic kidney disease (CKD), the accumulation of uremic toxins (UTs) promotes neuroinflammation and disrupts the blood–brain barrier. The results of animal studies suggest that UTs are involved in depression. Depression is strongly linked to neuroinflammation. In chronic kidney disease (CKD), the accumulation of uremic toxins (UTs) promotes neuroinflammation and disrupts the blood–brain barrier. The results of animal studies suggest that UTs are involved in depression. WHAT THIS STUDY ADDS This study of a large cohort of non-dialysed patients with CKD is the first to show an independent effect of phenylacetylglutamine on the progression of depressive symptoms and an influence of indoxyl sulphate and indole-3-acetic acid on symptom severity. This study of a large cohort of non-dialysed patients with CKD is the first to show an independent effect of phenylacetylglutamine on the progression of depressive symptoms and an influence of indoxyl sulphate and indole-3-acetic acid on symptom severity. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Our findings highlight the need for further research on the mechanisms underlying the influence of UTs on CKD-related depression and on the development of strategies that target UTs in the management of depressive symptoms. Our findings highlight the need for further research on the mechanisms underlying the influence of UTs on CKD-related depression and on the development of strategies that target UTs in the management of depressive symptoms. Introduction Depression is one of the most prevalent neuropsychiatric disorders in people with chronic kidney disease (CKD). Depending on the depression measurement scale used, the prevalence in the literature ranges from 26% to 50%—substantially higher than the value of 5% reported for the general population by the WHO.1 2 The high depression burden in CKD is due to the condition’s association with poor treatment adherence, poor quality of life, and an elevated risk of complications (including cardiovascular (CV) events, acute kidney injury and hospital admission) and death.3 The pathophysiology of depression involves several interconnected mechanisms: monoamine depletion, dysregulation of the hypothalamic–pituitary–adrenal axis, chronic inflammation (with immune dysfunction and elevated cytokine production), and impairment of the blood–brain—barrier (BBB).4 5 No one model can fully explain the complexity of depression in CKD. The association between CKD severity and depression suggests uremic toxins (UTs, ie, metabolites normally excreted by the kidneys and which accumulated in CKD) have a contributory role. Many UTs are generated by microbial metabolic activity in the intestine.6 In CKD, gut dysbiosis reduces beneficial bacteria and favours proteolytic species, leading to increase UT production.7 8 Furthermore, dysbiosis increases intestinal barrier permeability; this promotes UTs translocation into the circulation and exacerbates systemic inflammation and neuroinflammation.9 One of the major families of UTs is derived from tryptophan, an essential amino acid metabolised through several pathways (online supplemental figure S1). The kynurenine pathway produces nicotinamide, which regulates inflammation. The metabolites of nicotinamide act as glutamatergic modulators and probably contribute to mood disorders.10 The serotonin pathway generates serotonin and melatonin, while the indole pathway produces indole-3-acetic acid (IAA) and indoxyl sulfate (IS). IAA and IS are the most intensively studied UTs. In murine models, IS is associated with low serotonin levels in the brain and is implicated in disruption of the BBB.11 12 Other UTs (such as the choline derivative trimethylamine N-oxide (TMAO)) activate astrocytes and microglia, leading to neuroinflammation.13 Furthermore, the phenylalanine derivatives p-cresyl sulfate (pCS) and phenylacetylglutamine (PAG) have both demonstrated potential neurotoxic effects, including increased neuroinflammation, behavioural changes in mice and specific associations with cognitive impairment.14 15 Several studies (primarily in mouse models) suggest that UTs have a role in depression. However, to our knowledge, the relationship between UTs and depressive symptoms has not been studied in non-dialysed patients with CKD. The objective of this longitudinal study was to evaluate this association in a large, well-characterised cohort of non-dialysed adults with CKD. We hypothesised that independently of sociodemographic factors, CV risk factors and kidney function, higher serum UT levels would be associated with more severe depressive symptoms. This study aimed to provide new physiological insights into the link between CKD and depressive symptoms. Methods Population Between July 2013 and April 2016, 3033 patients with CKD stage 2–5 were included in the Chronic Kidney Disease-Renal Epidemiology and Information Network (CKD-REIN) cohort (ClinicalTrials.gov identifier: NCT03381950) by 40 randomly selected nephrology facilities in France.16 Patients with a history of kidney replacement therapy (KRT, ie, maintenance dialysis or kidney transplantation) were not included in the study. The patients had to be aged 18 or over and had to be able to give their informed consent. The study design involved the collection of data by clinical research associates (CRAs) at a baseline visit and then every year for 5 years of follow-up or up to 6 months after the initiation of KRT. For the purposes of the present analysis, we excluded patients lacking data on serum UT levels or on the depressive symptom burden (measured on the Center for Epidemiologic Studies Depression (CESD) scale) during the 5-year follow-up period. We also excluded patients whose CESD scores were measured solely after the initiation of KRT (figure 1). Flowchart of the study. CESD, Center for Epidemiologic Studies Depression; CKD-REIN, Chronic Kidney Disease-Renal Epidemiology and Information Network; UTs, uremic toxins. Available data CRAs obtained baseline and yearly follow-up data through medical record reviews, patient interviews and self-completed questionnaires. Screening procedures identified any history of diabetes, dyslipidaemia, current tobacco use, cerebrovascular or CV disease and past or current cognitive disease and/or depression. The sociodemographic variables included age, sex and living situation (alone or not). Global cognitive function was evaluated using the Mini-Mental State Examination (MMSE). Height and weight were recorded, in order to calculate the body mass index. Medication use during the 3 months prior to enrolment was documented by CRAs, using a standardised electronic case report form linked to the international Anatomical Therapeutic and Chemical thesaurus. Blood and urine tests routinely recommended by the French health authorities for the management of CKD were performed for all participants in their usual medical laboratories. The glomerular filtration rate (GFR) was calculated using the 2009 Chronic Kidney Disease Epidemiology Collaboration creatinine equation. The urinary albumin-to-creatinine ratio (uACR) was determined either directly or estimated from proteinuria measurements. More details of data collection are given in online supplemental Table S1. Exposure: UT assays All CKD-REIN samples were stored at −80 °C at the Biobanque de Picardie, Biological Resource Center (Amiens, France; BRIF number: BB-0033–00 017) and shipped frozen to Paris for analysis. Total and free fractions of PAG, pCS, TMAO, kynurenic acid, kynurenine, IAA, IS and two of the latter’s precursors (tryptophan and phenylalanine) were assayed using a validated liquid chromatography-tandem mass spectrometry technique. TMAO, PAG and phenylalanine are small, water-soluble molecules, while the other UTs are predominantly protein-bound. For more details, see the online supplementary material (online supplemental Table S1).17 Assessment of depressive symptoms We used the CESD scale to evaluate changes over time in depressive symptoms.18 The CESD is a validated screening instrument for depressive symptoms and does not allow for a clinical diagnosis of depression. It should, therefore, be interpreted as an indicator of symptom burden rather than diagnostic depression status. The original CESD consists of 20 items, each of which is rated from 0 (‘rarely or never’, ie, less than one day in the past week) to 3 (‘most of the time’, ie, 5–7 days per week). In the CKD-REIN cohort, a shortened 10-item version of the CESD (yielding a total score out of 30) was administered, as it is a validated instrument that reduces respondent burden in longitudinal studies, particularly in the context of multiple concomitant questionnaires.19 A higher CESD score reflects a greater burden of depressive symptoms. CESD scores were collected at baseline and at the 1-year, 2-year, 3-year and 5-year follow-up visits. The CESD has been widely used in patients with CKD to assess depressive symptom burden and has shown acceptable validity in this population. The shortened 10-item version has been validated against the full scale and demonstrates good psychometric properties.20 For the purpose of the present analysis, only CESD measurements obtained prior to KRT initiation were considered. Measurements collected after KRT initiation were excluded, and thus all analyses were restricted to the non-dialysis CKD phase. Statistical analysis We first compared the patients’ characteristics at baseline as a function of their baseline eGFR category (<30, 30–44 and ≥45 mL/min/1.73 m²). Continuous variables were analysed in an analysis of variance, while categorical variables were evaluated using the χ2 test. We considered the free fraction for pCS, IAA, kynurenic acid, kynurenine and IS (given that these UTs are mainly bound to albumin) and the total fraction for TMAO and PAG. With regard to the two UT precursors, we considered the free fraction for tryptophan (which is also mainly bound to albumin) and the total fraction for phenylalanine. To graphically illustrate the evolution of depressive symptoms over time, CESD scores were modelled using an unadjusted linear mixed-effects model with random intercepts and slopes. To assess the relationship between baseline UT levels and the mean changes over time in the CESD scores, we applied linear, mixed-effects models with random intercepts and slopes (the ‘lmer’ function and the lme4 package in R software (V.2025.05.0+496 (R Foundation for Statistical Computing, Vienna, Austria)). A linear time effect was retained as it provided a better fit than models using restricted cubic splines. CESD scores over time were modelled as a function of baseline UT levels, time and their interaction term, allowing estimation of the effect of UT levels on the rate of change in CESD scores. Models were progressively adjusted according to covariates selected based on a literature review. Model 1 included sociodemographic variables (age, sex and the MMSE score). Model 2 additionally included CV risk factors (systolic blood pressure in 10 mm Hg increments, diabetes, dyslipidaemia, obesity and current smoking) and mental health variables (history of depression and antidepressant intake at baseline). Model 3 further adjusted for kidney function markers (eGFR, expressed per 5 mL/min/1.73 m² decrement and uACR). Finally, model 4 additionally included uremic toxins previously described in preclinical or clinical studies as interacting with brain function or structure (free fractions of IS, indole-3-acetic acid, kynurenine, kynurenic acid, pCS and total fractions of TMAO and PAG). Among these, only toxins associated with CESD scores in unadjusted analyses (p<0.20) were retained in the final adjustment model. As described above, to assess the association between the baseline UT levels and the mean changes over time in the CESD score, we tested the interaction term between the baseline UT levels and time. Given that the serum level of each UT increases with the CKD stage, we also tested for interactions between the UTs and the eGFR. Moreover, we tested for collinearity between variables by calculating the variance inflation factor (the ‘vif’ function in the ‘car’ package in R). The homoscedasticity of the mixed model residuals was checked graphically. We used chained equations to perform multiple imputations of missing data (the ‘mice’ function in the ‘mice’ package in R). Thirty datasets were created with 30 iterations. All variables present in the linear mixed model were included in the imputation procedure. Effects were quoted with their corresponding 95% CI, and the threshold for statistical significance was set to p<0.05. All statistical analyses were performed with R software. Results Characteristics of the study population After the application of our exclusion criteria, 2165 patients were included in the analysis (figure 1, table 1). The median IQR age was 68 (60 to 76) and the mean (SD) eGFR was 34.9 (13.3) mL/min/1.73 m2. With regard to depressive symptoms, the median (IQR) CESD score was 7.0 (4.0 to 10.0) and 7.4% of the study population had been prescribed antidepressants at baseline. Out of the 2165 patients included in the study, 879 (40.6%) had an eGFR <30 mL/min/1.73 m2. The CESD score was significantly higher in patients with a lower eGFR (<30 mL/min/1.73 m2) than in patients with a higher eGFR. With the exception of kynurenic acid and phenylalanine, the median concentrations of UTs and UT precursors were significantly higher in patients with a lower eGFR than in patients with a higher eGFR (table 1). Table 1Baseline characteristics of the participants, according to the eGFR level (n=2165)NAOveralln=2165eGFR≥45 mL/min/1.73 m2n=478 (22.1%)eGFR45–30 mL/min/1.73 m2n=809 (37.4%)eGFR<30 mL/min/1.73 m2n=878 (40.6%)P*Sociodemographic factors Age, median (IQR)068.0 (60.0, 76.0)67.0 (58.3, 72.0)68.0 (61.0, 76.0)69.0 (61.0, 77.0) <0.001 Males, n (%)01436 (66.3)336 (70.3)542 (67.0)558 (63.6) 0.04 Educational level≥12 years, n (%)21 (1.0)788 (36.4)211 (44.1)289 (35.7)288 (32.8) 0.001 Living alone, n (%)187 (8.6)438 (20.2)74 (15.5)175 (21.6)189 (21.5)0.05Depression-related variables History of depression, n (%)15 (0.7)166 (7.7)43 (9.0)63 (7.8)60 (6.8)0.3 Short-form CESD score (out of 30), median (IQR)132 (6.1)7.0 (4.0, 10.0)6.0 (3.0, 10.0)6.0 (3.3, 10.0)7.0 (4.0, 11.0) 0.001 Antidepressants, n (%)17 (0.8)161 (7.4)31 (6.5)68 (8.4)62 (7.1)0.06Cardiovascular risk factors† and comorbidities Systolic blood pressure (mm Hg), mean (SD)20 (0.9)141.2 (19.8)138.3 (18.4)142.3 (19.8)141.7 (20.3) 0.001 Diabetes mellitus, n (%)5 (0.2)866 (40.0)181 (37.9)343 (42.4)342 (39.0)0.3 Dyslipidaemia, n (%)8 (0.4)1586 (73.3)333 (69.7)598 (73.9)655 (74.6)0.3 Obesity, n (%)38 (1.8)758 (35.0)140 (29.3)300 (37.1)318 (36.2) 0.01 Current smoking, n (%)13 (0.6)252 (11.6)57 (11.9)93 (11.5)102 (11.6)0.8 MMSE score (out of 30), median (IQR)17 (0.8)28.0 (25.0, 29.0)28.0 (26.0, 29.0)28.0 (25.0, 29.0)28.0 (25.0, 29.0) <0.001 Polymedication (≥5 prescription medications daily), n (%)9 (0.4)1703 (78.7)337 (70.5)612 (75.6)754 (85.9) <0.001 Standard laboratory variables eGFR (mL/min/1.73 m2), mean (SD)034.9 (13.3)54.0 (7.6)37.2 (4.3)22.3 (4.9) <0.001 uACR, mean (SD)231 (10.7) <0.001 Normal: <3 mg/mmol 552 (25.5)206 (43.1)228 (28.2)118 (13.4) Moderate elevation: 3–30 mg/mmol687 (31.7)143 (29.9)262 (32.4)282 (32.1) Severe elevation: >30 mg/mmol731 (33.8)89 (18.6)235 (29.0)407 (46.4) Uremic toxins Serum free tryptophan level (mg/L), median (IQR)00.9 (0.7, 1.2)0.9 (0.7, 1.1)0.9 (0.7, 1.2)0.9 (0.7, 1.2) 0.005 Serum free kynurenine level (mg/L), median (IQR)00.1 (0.1, 0.2)0.1 (0.07, 0.1)0.1 (0.09, 0.2)0.2 (0.1, 0.2) <0.001 Serum free kynurenic acid level (mg/L), median (IQR)00.01 (0.01, 0.01)0.01 (0.01, 0.01)0.01 (0.01, 0.01)0.01 (0.01, 0.01)0.9 Serum free indoxyl sulfate level (mg/L), median (IQR)00.05 (0.03, 0.09)0.03 (0.02, 0.05)0.04 (0.03, 0.07)0.08 (0.05, 0.2) <0.001 Serum free indole-3-acetic acid level (mg/L), median (IQR)00.03 (0.02, 0.05)0.03 (0.02, 0.04)0.03 (0.02, 0.05)0.04 (0.03, 0.06) <0.001 Serum total phenylalanine level (mg/L), median (IQR)016.9 (14.3, 20.3)17.1 (14.2, 20.6)16.7 (14.4, 20.0)17.0 (14.3, 20.6)0.8 Serum total phenylacetylglutamine level (mg/L), median (IQR)02.1 (1.2, 3.5)1.3 (0.8, 2.0)1.9 (1.1, 3.0)3.0 (1.8, 5.1) <0.001 Serum free p-cresyl sulfate level (mg/L), median (IQR)00.2 (0.1, 0.4)0.1 (0.05, 0.2)0.2 (0.07, 0.3)0.3 (0.2, 0.6) <0.001 Serum total TMAO level (mg/L), median (IQR)01.3 (0.7, 2.3)0.7 (0.5, 1.3)1.1 (0.7, 1.8)1.9 (1.2, 3.1) <0.001 *Student’s test for comparisons of means, the χ2 test for comparisons of percentages and the Wilcoxon test for comparisons of category distributions.†Hypertension was defined as a history of hypertension or the use of blood pressure-lowering medication. Diabetes mellitus was defined as a history of diabetes, antidiabetic medication use, a glycosylated haemoglobin level≥6.5%, a fasting glycaemia value≥7 mmol/L, or a non-fasting glycaemia value≥11 mmol/L. Dyslipidaemia was defined as a history of dyslipidaemia or the use of lipid-lowering medication. Obesity was defined as a body mass index≥30 kg/m2. Current smoking was defined as at least one cigarette per day.‡Psychoactive drug intake was defined as taking at least one antidepressant, anxiolytic, or antipsychotic.§Cerebrovascular disease was defined as a history of stroke, transient ischaemic attack or brain haemorrhage.CESD, Centre for Epidemiologic Studies Depression scale; CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; IQR, interquartile range; MMSE, Mini Mental State Examination; NA, not available; SD, standard deviation; TMAO, trimethylamine-N-oxide; uACR, urine albumin/creatinine ratio. Baseline characteristics of the participants, according to the eGFR level (n=2165) Student’s test for comparisons of means, the χ2 test for comparisons of percentages and the Wilcoxon test for comparisons of category distributions. Hypertension was defined as a history of hypertension or the use of blood pressure-lowering medication. Diabetes mellitus was defined as a history of diabetes, antidiabetic medication use, a glycosylated haemoglobin level≥6.5%, a fasting glycaemia value≥7 mmol/L, or a non-fasting glycaemia value≥11 mmol/L. Dyslipidaemia was defined as a history of dyslipidaemia or the use of lipid-lowering medication. Obesity was defined as a body mass index≥30 kg/m2. Current smoking was defined as at least one cigarette per day. Psychoactive drug intake was defined as taking at least one antidepressant, anxiolytic, or antipsychotic. Cerebrovascular disease was defined as a history of stroke, transient ischaemic attack or brain haemorrhage. CESD, Centre for Epidemiologic Studies Depression scale; CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; IQR, interquartile range; MMSE, Mini Mental State Examination; NA, not available; SD, standard deviation; TMAO, trimethylamine-N-oxide; uACR, urine albumin/creatinine ratio. Association between baseline UT levels and the mean change over time in the CESD score Over a mean follow-up period of 4.0 (1.5) years, the median number of CESD assessments per patient was 3.0 (2.0 to 4.0). In an unadjusted linear mixed model, we observed a slight increase in the CESD score over time (mean increase: 0.15 points (95% CI 0.09 to 0.21) (online supplemental Table S2). In the fully adjusted linear mixed-effects model, the CESD score increased by 0.11 (95% CI 0.04 to 0.18) points per year (figure 2). The baseline serum PAG level modified this rate of increase over time, as indicated by a significant interaction between the serum PAG level and time (p=0.02). Specifically, a two-fold increase in PAG was associated with an additional increase in the CESD slope of 0.06 (95% CI 0.01 to 0.18) points per year (table 2). Predict change in the mean CESD score over the 5-year follow-up (n= 2165). CESD, Center for Epidemiologic Studies Depression. Table 2Uremic toxins (concentrations considered as continuous variables) associated with the mean change in CESD score in a linear mixed model (n=2165)Adjusted effectModel 1Adjusted effectModel 2Adjusted effectModel 3Adjusted effectModel 4Effect (95% CI)PEffect (95% CI)PEffect (95% CI)PEffect (95% CI)Plog2 kynurenic acid−1.13 (−2.95 to 0.68)0.2−1.04 (−2.79 to 0.71)0.2−1.09 (−2.83 to 0.66)0.2−1.26 (−3.00 to 0.48)0.2log2 kynurenine0.50 (0.24 to 0.77) <0.001 0.47 (0.21 to 0.73) <0.001 0.21 (−0.07 to 0.49)0.10.05 (−0.25 to 0.35)0.7log2 indoxyl sulphate0.41 (0.27 to 0.55) <0.001 0.38 (0.25 to 0.52) <0.001 0.29 (0.14 to 0.45) <0.001 0.22 (0.02 to 0.41) 0.03 log2 indole-3-acetic acid0.50 (0.30 to 0.71) <0.001 0.49 (0.29 to 0.69) <0.001 0.40 (0.19 to 0.61) <0.001 0.30 (0.07 to 0.52) 0.01 log2 phenylacetylglutamine Interaction Interaction Interaction Interaction log2 p-cresyl sulphate0.22 (0.10 to 0.33) <0.001 0.21 (0.10 to 0.32) <0.001 0.12 (0.00 to 0.24)0.05−0.12 (−0.29 to 0.06)0.2log2 p-TMAO0.25 (0.10 to 0.21) <0.001 0.24 (0.09 to 0.39) 0.002 0.10 (−0.06, 0.27)0.2−0.009 (−0.19 to 0.17)0.9 Interaction log2 phenylacetylglutamine x Time * CESD change per year0.11 (0.04 to 0.18) 0.002 0.11 (0.04 to 0.18) 0.002 0.11 (0.04 to 0.18) 0.002 0.11 (0.04 to 0.18) 0.002 * Additional CESD change for a doubling in the PAG level0.06 (0.01 to 0.11) 0.02 0.06 (0.01 to 0.11) 0.02 0.06 (0.01 to 0.11) 0.02 0.06 (0.01 to 0.11) 0.02 For all models, multiple imputation by chained equations was applied to the dataset. Model 1 was adjusted for log2 of the concentration of the uremic toxin of interest, age (per 10-year increment), sex and MMSE score.Model 2 was adjusted for log2 of the concentration of the uremic toxin of interest, age (per 10-year increment), sex and MMSE score, systolic blood pressure (per 10 mm Hg increment), diabetes mellitus, dyslipidaemia, obesity, smoking, history of depression and antidepressant intake at baseline. Model 3 was adjusted for log2 of the concentration of the uremic toxin of interest, age (per 10-year increment), sex and MMSE score, systolic blood pressure (per 10 mm Hg increment), diabetes mellitus, dyslipidaemia, obesity, smoking, history of depression, antidepressant intake at baseline, eGFR (per 5 mL/min/1.73 m2 decrease) and urine albumin/creatinine ratio. Model 4 was adjusted for log2 of the concentration of the uremic toxin of interest, age (per 10-year increment), sex, and MMSE score, systolic blood pressure (per 10 mm Hg increment), diabetes mellitus, dyslipidaemia, obesity, smoking, history of depression, antidepressant intake at baseline, eGFR (per 5 mL/min/1.73 m2 decrease), urine albumin/creatinine ratio and UT associated with CESD score in crude model with p value<0.2. For kynurenic acid: adjustment for log2(kynurenine), log2(IS), log2(IAA), log2(PAG), log2(pCS), log2(TMAO). For kynurenine: adjustment for log2(IS), log2(IAA), log2(PAG), log2(pCS), log2(TMAO). For indoxyl sulphate: adjustment for log2(kynurenine), log2(IAA), log2(PAG), log2(pCS), log2(TMAO). For indole-3-acetic acid: adjustment for log2(kynurenine), log2(IS), log2(PAG), log2(pCS), log2(TMAO). For phenylacetylglutamine: adjustment for log2(kynurenine), log2(IAA), log2(IS), log2(pCS), log2(TMAO). For p-cresyl sulphate: adjustment for log2(kynurenine), log2(IAA), log2(PAG), log2(IS), log2(TMAO). For TMAO: adjustment for log2(kynurenine), log2(IAA), log2(PAG), log2(IS), log2(pCS).CESD, Centre for Epidemiologic Studies Depression scale; eGFR, estimated glomerular filtration rate; IAA, indole-3-acetic acid; IS, indoxyl sulphate; MMSE, Mini Mental State Examination; PAG, phenylacetylglutamine; pCS, p-cresyl sulphate; TMAO, trimethylamine-N-oxide; uACR, urine albumin/creatinine ratio; UT, uremic toxin. Uremic toxins (concentrations considered as continuous variables) associated with the mean change in CESD score in a linear mixed model (n=2165) For all models, multiple imputation by chained equations was applied to the dataset. Model 1 was adjusted for log2 of the concentration of the uremic toxin of interest, age (per 10-year increment), sex and MMSE score. Model 2 was adjusted for log2 of the concentration of the uremic toxin of interest, age (per 10-year increment), sex and MMSE score, systolic blood pressure (per 10 mm Hg increment), diabetes mellitus, dyslipidaemia, obesity, smoking, history of depression and antidepressant intake at baseline. Model 3 was adjusted for log2 of the concentration of the uremic toxin of interest, age (per 10-year increment), sex and MMSE score, systolic blood pressure (per 10 mm Hg increment), diabetes mellitus, dyslipidaemia, obesity, smoking, history of depression, antidepressant intake at baseline, eGFR (per 5 mL/min/1.73 m2 decrease) and urine albumin/creatinine ratio. Model 4 was adjusted for log2 of the concentration of the uremic toxin of interest, age (per 10-year increment), sex, and MMSE score, systolic blood pressure (per 10 mm Hg increment), diabetes mellitus, dyslipidaemia, obesity, smoking, history of depression, antidepressant intake at baseline, eGFR (per 5 mL/min/1.73 m2 decrease), urine albumin/creatinine ratio and UT associated with CESD score in crude model with p value<0.2. For kynurenic acid: adjustment for log2(kynurenine), log2(IS), log2(IAA), log2(PAG), log2(pCS), log2(TMAO). For kynurenine: adjustment for log2(IS), log2(IAA), log2(PAG), log2(pCS), log2(TMAO). For indoxyl sulphate: adjustment for log2(kynurenine), log2(IAA), log2(PAG), log2(pCS), log2(TMAO). For indole-3-acetic acid: adjustment for log2(kynurenine), log2(IS), log2(PAG), log2(pCS), log2(TMAO). For phenylacetylglutamine: adjustment for log2(kynurenine), log2(IAA), log2(IS), log2(pCS), log2(TMAO). For p-cresyl sulphate: adjustment for log2(kynurenine), log2(IAA), log2(PAG), log2(IS), log2(TMAO). For TMAO: adjustment for log2(kynurenine), log2(IAA), log2(PAG), log2(IS), log2(pCS). CESD, Centre for Epidemiologic Studies Depression scale; eGFR, estimated glomerular filtration rate; IAA, indole-3-acetic acid; IS, indoxyl sulphate; MMSE, Mini Mental State Examination; PAG, phenylacetylglutamine; pCS, p-cresyl sulphate; TMAO, trimethylamine-N-oxide; uACR, urine albumin/creatinine ratio; UT, uremic toxin. For the other UTs (kynurenic acid, kynurenine, IS, IAA, TMAO and pCS), no association was found between baseline serum levels and the trajectory of mean CESD over time. However, some significant differences in mean CESD scores were observed (table 2). The kynurenic acid level was not associated with mean CESD score—even in the unadjusted model. For pCS, kynurenine and TMAO, a doubling of the UT level was associated with a higher mean CESD score after adjustment for age, sex, MMSE score, CV risk factors, history of depression and antidepressant use (model 2), with increments of 0.21 (95% CI 0.10 to 0.32) for pCS, 0.47 (95% CI 0.21 to 0.73) for kynurenine and 0.26 (95% CI 0.11 to 0.41) for TMAO. However, these associations were no longer significant after additional adjustment for eGFR and uACR (model 3). For IS and IAA, a doubling of the UT level was associated with a higher mean CESD score in the fully adjusted model; the increment was 0.22 (95% CI 0.02 to 0.41) for IS and 0.30 (95% CI 0.07 to 0.52) for IAA. A significant interaction with the eGFR was not observed for any of the UTs. Discussion To the best of our knowledge, the present study is the first to have evidenced an association between the serum PAG level and changes over time in depressive symptoms in non-dialysed, non-transplanted patients with CKD. This association remained statistically significant after adjustment for sociodemographic factors, CV risk factors, antidepressant use, a history of depression, kidney function and the levels of UTs known to interact with brain function or structure. Although no association was found between baseline IS or IAA levels and the change over time in the CESD score, a doubling in the serum IS and IAA levels was significantly associated with the mean CESD score in the fully adjusted model. In the present study, a doubling in the PAG concentration was associated with a ~50% steeper CESD slope. This finding suggests that PAG contributes specifically to CKD-related neuropsychiatric complications and is in line with our previous observation of an association between a higher PAG level and lower cognitive level.15 Several studies have evidenced an association between the PAG level and CV or metabolic disease. Consideration of the metabolic pathway indicates that PAG (1) enhances platelet responsiveness by interacting with adrenergic receptors and (2) increases inflammation by upregulating pro-inflammatory markers and oxidative stress.21 Although the specific effect of PAG on brain structures remains poorly understood, the compound’s pro-inflammatory properties might significantly influence the development of neuropsychiatric disorders like depression. Indeed, many studies have shown that inflammation has a role in the development of depression.4 A recent study reported that PAG binds strongly to interleukin (IL)−1β and glycogen synthase kinase−3B; these components of the IL-17, Toll-like receptor and phosphatidylinositol 3’-kinase-protein kinase B intracellular signalling pathways are involved in the inflammation underlying depression.22 Our present results showed that the serum PAG level was associated with an increase over time in depressive symptoms, independently of kidney function. Further studies are needed to better understand the mechanisms by which the PAG level rises in CKD and to determine appropriate targets for reducing the compound’s toxicity. Although the baseline serum levels of IS or IAA were not associated with the mean CESD trajectory, they were significantly associated with the mean CESD level in the fully adjusted model. IS is currently the most extensively studied indole UT. It induces a reduction in serotonin production, activates the aryl hydrocarbon receptor (AhR), and thus leads to behavioural changes in rats and mice.10 The accumulation of IS also leads to brain inflammation (a classical element of treatment-resistant depression) via activation of the AhR and nuclear factor-kappa B pathways. Exposure to high levels of IS increases astrocyte and microglial activation and the release of pro-inflammatory cytokines. These effects are associated with elevated BBB permeability, which suggests that tryptophan-derived UTs are major contributors to depression in CKD.23 In addition to its association with depression, the IS concentration has also been linked to the severity of anxiety in non-CKD populations.24 Regarding kynurenine and TMAO, the associations with the mean CESD score became non-significant after eGFR was included in the model; this suggests that kidney function mediates or confounds the relationship observed here. In vivo studies in mice show that plasma and brain levels of TMAO increase with age and exert age-dependent effects on pro-inflammatory cytokine production, astrocyte activation and BBB integrity.13 25 The kynurenine pathway is notably activated by inflammation and diverts tryptophan away from serotonin synthesis towards the production of neuroactive metabolites that might produce depressive symptoms. Kynurenine crosses the BBB and is metabolised in the brain into compounds such as kynurenic acid. Elevated levels of kynurenic acid and the dysregulation of kynurenic acid homeostasis can impair glutamatergic and dopaminergic signalling, which may contribute to cognitive and motivational impairments. Consequently, kynurenine metabolites are viewed as key factors in the pathophysiology of depression.26 27 In contrast to the literature data, our present results did not evidence a significant association between the tryptophan–kynurenine pathway and depression; hence, the disease mechanism(s) underlying depression might be different in people with CKD. Although UTs arise from different metabolic pathways, they all contribute to neuroinflammation by stimulating pro-inflammatory cytokine production and reducing BBB permeability. In turn, the reduced permeability facilitates monocyte infiltration and microglial activation, impairs hippocampal neurogenesis and disrupts synaptic remodelling across interconnected brain regions.28 Despite these shared mechanisms, PAG was the only UT independently associated with the progression of depressive symptoms, even after adjustment for kidney function and other UTs known to contribute to neuroinflammation. The indole tryptophan pathway (IS and IAA) was associated with the mean CESD score, but not its change over time, while kynurenine-derived metabolites showed no association after adjustment. These differences among tryptophan-derived metabolites raise questions about which metabolic route are preferentially activated in CKD and why. More broadly, uncertainties remain regarding UT neurotoxicity, including potential interactions and the timing of UT-induced brain changes. Our results did not evidence collinearity or interactions between the studied UTs. More generally, the analysis of UTs might provide insights into the kidney–gut–brain axis and the latter’s role in depressive symptoms. Finally, we can hypothesise that strategies for reducing UT production (through dietary means or by administering probiotics) and/or promoting UT elimination (eg, using chelators) would help to reduce UT levels and thus perhaps lessen depressive symptoms.29 Although we found an association between the mean CESD score and levels of PAG, IS and IAA in model 4, the effect size was modest. This finding should be interpreted in a context where many other risk factors strongly associated with depression were present. UTs likely represent only one component of this complex puzzle and cannot fully explain depressive symptoms in people with CKD. In this study, depressive symptoms and initiation of antidepressant treatment were self-reported and not evaluated by clinicians. This aspect is particularly relevant in CKD, where overlapping symptoms (fatigue, sleep disturbances, sexual dysfunction and cognitive impairment) complicate the distinction between depressive symptoms and CKD-related manifestations. Two CESD items relate to fatigue and sleep disturbances, which are common in CKD. This overlap might explain the low prevalence of antidepressant prescription observed in our cohort and in other CKD populations, despite the higher reported prevalence of depression.3 In our cohort, the median CESD score at baseline was 7/30, whereas the threshold for suspected depression is 10/30; hence, the depressive symptom burden was relevant in our study population.30 The present study had several limitations. First, it was subjected to selection bias inherent to cohort studies, as depressed patients are less likely to be included, attend follow-up visits or complete self-questionnaires. These biases might have influenced our findings by (1) excluding the most depressed individuals at enrolment and (2) reducing the number of patients with higher depressive symptom burden, and multiple CESD assessments, resulting in a small mean change over time. Second, patients with more advanced CKD (who usually have higher UT levels and more intense depressive symptoms) were more likely to leave the study early, due to dialysis, transplantation or death, leading to fewer CESD measurements. Consequently, missing data were more likely, as follow-up ended once KRT was initiated. These selection effects might have contributed to the weak associations between UTs and the CESD score, and (given the very limited variation in the CESD trajectory) the lack of an effect of IS, IAA, pCS, TMAO and kynurenine on CESD score changes. Third, the 5-year follow-up period may have been insufficient to detect the onset of depressive symptoms. Fourth, the CESD was the only measure used to assess depressive symptoms, and no structured clinical interview was available to confirm a diagnosis of depression. As a screening instrument, the CESD does not allow for clinical diagnosis and may be influenced by somatic symptoms, particularly in patients with CKD, which could lead to an underestimation or overestimation of depressive symptom burden. In addition, a shortened 10-item version of the CESD was used instead of the full 20-item scale to reduce respondent burden in the context of repeated assessments. Although this abbreviated version has been validated and shown to correlate strongly with the original scale, it may provide less granular measurement of depressive symptoms. Fifth, dietary data were not collected in the CKD-REIN study, preventing adjustment for this factor. Finally, this exploratory study aimed to identify mechanisms underlying the association between UT levels and depressive symptoms; further studies are needed to determine whether this relationship is causal. The present study also had several strengths. First, the CKD-REIN cohort’s extensive datasets allowed assessment of a wide range of variables and adjustment for numerous potential confounders. Even after applying exclusion criteria, 2165 patients were included. Second, all UT assays were centralised and performed using a validated, robust, ultra-high-performance liquid chromatography–tandem mass spectrometry technique. Third, the UTs analysed were known to be associated with complications of CKD in preclinical studies. Fourth, the CESD scale was scored regularly during follow-up, enabling random-slope mixed models and inclusion of CESD data from all participants, not only those with at least two measurements. Conclusion To the best of our knowledge, the present study of a large cohort of non-dialysed, non-transplanted patients with CKD is the first to have highlighted an independent association between the serum PAG level and a longitudinal change in depressive symptoms (as assessed on the CESD scale). Although baseline IS and IAA levels were not associated with changes in CESD over time, they were significantly related to the mean CESD score. These findings constitute a strong rationale for further research in the field of CKD and provide a better understanding of the potential role of UTs in depressive symptoms. Future studies are needed to further clarify the role of uremic toxins in the development and progression of depressive symptoms in CKD and to determine whether they may have a potential role within a broader network of biological and clinical factors, contributing to depressive symptom burden, and how they interact with other metabolic, inflammatory and psychosocial determinants within this framework. Finally, preclinical studies of PAG and tryptophan metabolism are needed to underpin the development of new strategies that could target UTs in the management of depressive symptoms. Supplementary material Acknowledgements We thank the CKD-REIN study coordination staff for their efforts in setting up the CKD-REIN cohort (Marie Metzger, Oriane Lambert, Céline Lange, Heliz Argan, Emilie Moutard, Kélamaé Oulai and all the clinical research associates), as well as the staff at the CKD-REIN clinical sites and the investigators. We also thank the staff at the participating biological resource centres: the Biobanque de Picardie (BB-0033-00017), NeuroBioTec (BB-0033-00046), Centre de ressources biologiques (CRB)-Centre Hospitalier Universitaire de Nantes Hôtel Dieu (BB-0033-00040), CRB-Centre Hospitalier Universitaire Grenoble Alpes (BB-0033-00069), CRB-Centre Hospitalier Régional Universitaire de Nancy (BB-0033-00035), Service de Néphrologie, Centre Hospitalier de Perpignan, the Plateforme de Ressources Biologiques-Hôpital Henri Mondor (BB-0033-00021), the Centre d’Investigation Clinique Plurithématique CIC-1435, Plateforme de Ressources Biologiques-Hôpital européen Georges-Pompidou (BB-0033-00063), L’Etablissement Français du sang (EFS) Hauts de France – Normandie (Site de Bois-Guillaume, Site de Loos-Eurasanté), EFS Nouvelle Aquitaine (site Pellegrin), EFS Ile de France (Site Avicenne), EFS Occitanie (Site de Toulouse), EFS Grand-Est (Site de Colmar, Site de Metz), EFS PACA-Corse (Site de Marseille) (see online supplemental table S3). Lastly, we thank Dr David Fraser (Biotech Communication SARL, Ploudalmézeau, France) for copyediting assistance. Footnotes Funding: CKD-REIN is funded by the Agence Nationale de la Recherche through the 2010 'Cohortes-Investissements d’Avenir' program (number ANR-IA-COH-2012/3731), a subsidy from the Ministry of Higher Education and Research for public service charges (number 2023-1322 of 29 December 2023 of the Finance Ministry's budget for 2024), and by the 2010 national Programme Hospitalier de Recherche Clinique (number 2016-1348-DRC-FI-D). CKD-REIN is also supported through a public-private partnership with GlaxoSmithKline (GSK) since 2012, Boehringer Ingelheim France since 2022, Novo Nordisk since 2024, Fresenius Medical Care from 2012 to 2024, Vifor France from 2018 to 2023, Sanofi-Genzyme from 2012 to 2015, Baxter and Merck Sharp & Dohme-Chibret (MSD France) from 2012 to 2017, Amgen from 2012 to 2020, Lilly France from 2013 to 2018, Otsuka Pharmaceutical from 2015 to 2020, and AstraZeneca from 2018 to 2021. Inserm Transfert set up and has managed this partnership since 2011. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review or approval of the manuscript; and decision to submit the manuscript for publication. Provenance and peer review: Not commissioned; externally peer-reviewed. Patient consent for publication: Not applicable. Ethics approval: All legally required authorisations were obtained, including those from the Comité consultatif sur le traitement de l’information en matière de recherche dans le domaine de la santé (CCTIRS N°12.360), the Commission nationale de l’informatique et des libertés (CNIL N°DR-2012-469) and the Kremlin-Bicêtre Comité de protection des personnes (CPP N°IDRCB 2012-A00902-41). The CKD-REIN biological collection is registered in the management application of the COnservation D’Eléments du COrps Humain (CODECOCH N°-2012-1624). The Institut national de la santé et de la recherche médical (INSERM) institutional review board approved the study protocol (IRB00003888). ClinicalTrials.gov Identifier: NCT03381950. Participants gave informed consent to participate in the study before taking part. Data availability free text: The data that support the findings of this study are available upon reasonable request by contacting the CKD-REIN study coordination team at ckdrein@inserm.fr. Collaborators: The CKD-REIN Study Collaborators include members of the steering committee, staff at the coordinating centre and site investigators. Steering committee and study coordination: Natalia Alencar de Pinho, Dorothée Cannet, Christian Combe, Denis Fouque, Luc Frimat, Aghilès Hamroun, Yves-Edouard Herpe, Christian Jacquelinet, Maurice Laville, Sophie Liabeuf, Ziad A. Massy, Pascal Morel, Roberto Pecoits-Filho, Joost Schanstra, Bénédicte Stengel, Céline Lange, Oriane Lambert, Marie Metzger. CKD-REIN site investigators: Alsace: Professors T. Hannedouche and B. Moulin (CHU, Strasbourg), Dr A. Klein (CH Colmar) Aquitaine: Professor C. Combe (CHU, Bordeaux), Dr J.P. Bourdenx (Clinique St Augustin, Bordeaux), Dr A. Keller, Dr C. Delclaux (CH, Libourne), Dr B. Vendrely (Clinique St Martin, Pessac), Dr B. Deroure (Clinique Delay, Bayonne), Dr A. Lacraz (CH, Bayonne) Basse Normandie: Dr T. Lobbedez (CHU, Caen), Dr I. Landru (CH, Lisieux) Ile de France: Professor Z. Massy (CHU, Boulogne – Billancourt), Professor P. Lang (CHU, Créteil), Dr X. Belenfant (CH, Montreuil), Professor E. Thervet (CHU, Paris), Dr P. Urena (Clinique du Landy, St Ouen), Dr M. Delahousse (Hôpital Foch, Suresnes) Languedoc – Roussillon: Dr C. Vela (CH, Perpignan) Limousin: Professor M. Essig, Dr D. Clément (CHU, Limoges) Lorraine: Dr H. Sekhri, Dr M. Smati (CH, Epinal) Dr M. Jamali, Dr B. Hacq (Clinique Louis Pasteur, Essey-les-Nancy), Dr V. Panescu, Dr M. Bellou (Polyclinique de Gentilly, Nancy), Professor Luc Frimat (CHU, Vandœuvre-les-Nancy) Midi-Pyrénées: Professor N Kamar (CHU, Toulouse) Nord-Pas-de-Calais: Professors C. Noël and F. Glowacki (CHU, Lille), Dr N. Maisonneuve (CH, Valenciennes), Dr R. Azar (CH, Dunkerque), Dr M. Hoffmann (Hôpital privé La Louvière, Lille) Pays-de-la Loire: Professor M. Hourmant (CHU, Nantes), Dr A. Testa (Centre de dialyse, Rezé), Dr D. Besnier (CH, St Nazaire) Picardie: Professor G. Choukroun (CHU, Amiens), Dr G. Lambrey (CH, Beauvais) Provence-Alpes - Côte d’Azur: Professor S. Burtey (CHU, Marseille), Dr G. Lebrun (CH, Aix-en-Provence), Dr E. Magnant (Polyclinique du Parc Rambot, Aix-en-Provence) Rhône-Alpes: Professor M. Laville, Professor D. Fouque (CHU, Lyon-Sud) and L. Juillard (CHU Edouard Herriot, Lyon), Dr C. Chazot (Centre de rein artificiel Tassin Charcot, Ste Foy-les-Lyon), Professor P. Zaoui (CHU, Grenoble), Dr F. Kuentz (Centre de santé rénale, Grenoble). Contributor Information on behalf of CKD-REIN: Natalia Alencar de Pinho, Dorothée Cannet, Christian Combe, Denis Fouque, Luc Frimat, Aghilès Hamroun, Yves-Edouard Herpe, Christian Jacquelinet, Maurice Laville, Sophie Liabeuf, Ziad A. Massy, Pascal Morel, Roberto Pecoits-Filho, Joost Schanstra, Bénédicte Stengel, Céline Lange, Oriane Lambert, and Marie Metzger Data availability statement Data are available upon reasonable request. References Associated Data Supplementary Materials Data Availability Statement Data are available upon reasonable request.