Multiplatform urine metabolomics for non-invasive prediction of one-year renal function decline in kidney transplant recipients: a pilot study

Multiplatform urine metabolomics for non-invasive prediction of one-year renal function decline in kidney transplant recipients: a pilot study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Multiplatform urine metabolomics for non-invasive prediction of one-year renal function decline in kidney transplant recipients: a pilot study Arianna Cirillo, Guillaume Resimont, Justine Massias, Yann Guitton, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8620894/v1 This work is licensed under a CC BY 4.0 License Status: Under Revision Version 1 posted 14 You are reading this latest preprint version Abstract Introduction Kidney transplantation (KTx) provides the best therapeutic outcomes for patients with end-stage renal disease. However, long-term graft survival remains a major clinical challenge, and current biomarkers are insufficient to reliably predict post-transplant kidney function evolution. Identifying early predictors of renal function decline is therefore crucial to improve the monitoring and management of kidney transplant recipients (KTRs). Objectives This pilot study aimed to investigate whether multiplatform urine metabolomics could identify early predictive biomarkers of renal function decline between 3 and 12 months post-KTx. Methods A cohort of 56 French KTRs was recruited. Measured glomerular filtration rate (mGFR) was assessed at 3 (M3) and 12 (M12) months post-transplant, while urine samples were collected at M3. Patients were classified as “progressor” or “stable” based on a ≥ 7% decline or stability in mGFR over the 9-month follow-up period. Untargeted metabolomic profiling was performed on urine samples using complementary Nuclear Magnetic Resonance (NMR) and Mass Spectrometry (MS) platforms. Multivariate statistical analyses were then applied to identify metabolites associated with mGFR evolution. Results Multivariate modeling revealed putative urinary biomarkers associated with renal function trajectories. The strongest predictive performance was achieved using a combined model integrating both MS- and NMR-derived biomarkers, highlighting the complementarity of the two analytical approaches. Conclusion Despite being conducted on a relatively small cohort, this exploratory study demonstrates that urinary metabolomics, particularly when combining NMR and MS datasets, holds promise as a predictive tool for renal function evolution in kidney transplant recipients. These findings support further validation in larger, independent cohorts. metabolomics kidney transplant recipients NMR LC-MS multiplatform approach patient follow-up Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 1 Introduction Chronic kidney disease (CKD) is a common disease characterized by chronic alteration and progressive decline in kidney function that affects 5 to 10% of the general population 1 . In its most severe form, called end-stage renal disease (ESRD), dialysis or kidney transplantation (KTx) are required 2 . Nowadays, KTx represents the best treatment for patients with ESRD in terms of quality of life, morbidity and mortality 3 . Despite advancements, KTx is not devoid of risks, particularly concerning allograft dysfunction, which can result in compromised kidney function and graft rejection. Notably, 30–40% of patients experience graft loss within 10 years post-KTx 4 , 5 . In this regard, monitoring kidney function in kidney transplant recipients (KTRs) is essential and poses a critical concern during the follow-up period. Currently, the absence of biomarkers capable of predicting early kidney function loss post-transplantation is a notable challenge. In such a scenario, metabolomics emerges as a promising tool. Metabolomics, as part of the omics sciences, is based on the identification and quantification of small molecules, called “metabolites” (< 1500Da) that describe a biological system at a specific time-point 6 , 7 . These metabolites represent the downstream products of a series of biochemical events that occur in an organism under physiological or pathophysiological conditions. Nuclear magnetic resonance (NMR) and mass spectrometry (MS) stand as the primary analytical techniques utilized in metabolomics investigations. NMR platform allows non-targeted, non-destructive, and highly reproducible and robust sample analysis. On the other hand, MS holds a prominent place in metabolomics research due to its heightened sensitivity and its ability to detect and identify a multitude of metabolites within biofluids. 8 . Thanks to advancements in technical instruments and softwares, metabolomics is nowadays increasingly applied to detect disease profiles, including renal dysfunction, with several studies highlighting the link between biomarkers and kidney diseases 9 , 10 , 11 , 12 , 13 , 14 . Over the past decade, it has become a key tool in understanding CKD, providing insights into disease mechanisms and aiding the identification of diagnostic biomarkers. Many studies have focused on different aspects of CKD, such as disease stage 9 , 11 , 15 , 16 and models for estimating eGFR 15 , 17 . While CKD has been extensively studied, there is a notable gap in research on kidney function decline in post-KTx, with limited literature available in this area and no predictive model for GFR evolution in KTR. Furthermore, most metabolomics studies performed in the CKD field are based on estimated GFR, which is inherently less precise compared to measured GFR (mGFR), which represents the actual gold standard for the kidney function evaluation. For these reasons, we designed this exploratory pilot study to combine the strengths of NMR and MS platforms to analyze urine samples of a well-phenotyped cohort of KTRs, with the expectation of pinpointing early biomarkers that could predict the mid-term (at 1-year post-KTX) deterioration of kidney function; which has been measured and not estimated. For this purpose, we were able to take advantage of 3 months urine samples from KTRs prospectively collected during their protocol visits at 3 and 12 months post-KTx, during which GFR was measured. 2 Materials and Methods 2.1 Patient population and clinical data Fifty-six KTRs were included from April 2016 to November 2017. Urine samples were collected at 3 months post KTx and GFR was measured at 3 and 12 months after KTx in Bichat Hospital (Paris, France). The study was performed in accordance with relevant guidelines and regulations, and informed consent was obtained from all study participants. Patients were classified in two groups based on the evolution of measured GFR (mGFR) between 3 and 12 months after KTx. More specifically, mGFR was obtained by measuring the plasma clearance of 51 CrEDTA, which is considered as a reference method. We used the absolute values of mGFR (non-indexed for body surface area) and calculated the variation of GFR between 3 and 12 months (expressed in %). Knowing the relative change of mGFR (ΔGFR), we classified the patients as “Progressors” or “Stable”. “Progressors” were patients whose ΔGFR was declining of more than 7% and other were considered as “Stable”. The use of 7% as threshold in our stratification method is based on the concept of critical difference obtained at Bichat-Claude-Bernard Hospital were samples and clinical data were collected. The critical difference is based on the intra-individual coefficient of variation of GFR and can be defined as the smallest change in results of GFR which is not due to chance 18 . 2.2 Quality control (QC) samples Quality control (QC) samples were included to assess analytical reproducibility, batch effects, and system stability for both NMR and LC–MS analyses. 2.2.1 NMR For NMR analysis, two complementary QC strategies were employed. A pooled QC sample was generated by combining aliquots from all study participants, providing a representative overview of the cohort’s metabolic composition. As sample preparation and acquisition were conducted over two consecutive days, this pooled QC was prepared as two independent pools, each corresponding to one analytical batch. In addition, urine samples obtained from individuals external to the study cohort were included as external QCs to monitor instrument performance and inter-day stability independently of cohort-specific biological variability. 2.2.2 LC-MS For LC–MS analysis, QC samples consisted of a pooled urine sample generated from aliquots of all study participants. To further evaluate analytical performance across concentration ranges, diluted pooled QCs (2× and 4× dilutions) were also prepared and analyzed. 2.3 Sample preparation Metabolomics were performed on urine samples collected at 3 months in all patients. Urinary samples, corresponding to the second morning urine in fasting conditions. Urines were frozen without additives at − 80° C. 2.3.1. NMR Aliquots of 500µl of urine samples collected at 3-month were thawed on ice, supplemented with KF and prepared by following an optimized procedure for reduce inter-sample chemical-shift variations 19 . After this procedure, urine samples were supplemented with 200µl of deuterated phosphate buffer (DPB, pH 7.4), 100µl of a 5mM solution of maleic acid and 10µl of a 10 mg/ml TMSP D 2 O solution for NMR analysis. Urine pH and osmolality were measured before sample analysis. All QC samples were processed using the same protocol as study samples. 2.3.2. LC-MS An aliquot of 500µl of urine sample collected at 3-month was placed in 10kDa centrifugal filter (VWR, Fontenay-sous-Bois, France) and centrifuged at 13000g during 30min at 5°C. Once filtered, internal deuterated standards (leucine-5,5,5- d 3, L-tryptophan 2,3,3 d 3, indole-2,4,5,6,7- d 5-3-acetique acid et 1,14 tetradecanedoic- d 24 acid) were added to each sample and nitrogen blowdown was done. QC samples were prepared and processed in parallel using the same procedure. 2.4. Sample measurement 2.4.1. NMR All samples were recorded at 298 K on a Bruker Avance HD spectrometer operating at 700.17 MHz for the proton signal acquisition. The instrument was equipped with a TCI 5-mm cryoprobe with a Z-gradient. Maleic acid was used as the internal standard for quantification and trimethylsilyl-3-propionic acid- d 4 (TMSP) for the zero for the zero calibration. 1 H-NMR spectra were acquired using a 1D NOESY sequence with presaturation. The Noesypresat experiment used a RD-90°-T1-90°-Tm-90°-acquire sequence with a relaxation delay of 4 s, a mixing time (Tm) of 10 ms and a fixed T1 delay of 4 µs. Water suppression pulse was placed during the relaxation delay (RD). The number of transients is 64 (64K data points). QC samples were injected at regular intervals, every 10 samples, throughout the analytical sequence. The data were processed with the Bruker Topspin 4.0.8 software with a standard parameter set. Phase and baseline corrections were performed manually over the entire range of the spectra and the δ scale was calibrated to 0 ppm using the internal standard TMSP. System suitability and inter-batch instrumental drift was assessed by using QC samples. System performance was verified using the TMSP signal at 0.00 ppm based on chemical shift stability (± 0.01 ppm), line width (FWHM ≤ 1.2 Hz), and integral reproducibility (%CV ≤ 10%). 2.4.2. LC-MS All samples were analyzed on ultrahigh performance liquid chromatography with high-resolution mass spectrometry (UHLPC/MS) by following described method 20 for reversed phase(RP) UHPLC/MS. Tacking advantage of the MS2 capacities of the hybrid quadrupole-orbitrap (Q-Exactive TM) mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) QC samples (i.e.. pooled samples) were analyzed, in ESI positive and ESI negative modes, with three cycles of iterative Data Dependent MS 2 . The acquisition of the raw data was performed using a full scan mode within the m/z 65-1000 range at a resolving power of 70,000 at m/z 200,000 Da. Compound separation was performed using a Hypersil GOLD-C18 column (1.9 µm, 100 mm x 2.1 mm) from Thermo-Scientific (USA). The column temperature was set at 35°C. The mobile phases were composed of 0.1% of acetic acid in water (solvent A) and in acetonitrile (solvent B). Acetonitrile LC/HRMS grade CHROMASOLV™ LC-MS (Riedel-de Haën), Water LC/HRMS grade CHROMANORM® (VWR Chemicals). The applied gradient (A:B, v/v) was as follows: 95:5 from 0 to 2.4 min, 75:25 at 4.5 min, 25:75 at 11 min, 0:100 from 14 at 16.5 min and 95:5 from 19 to 25 min. The flow rate was set to 0.40 mL/min. The injection volume was 5 µL. All samples were analyzed in one batch without any stopping or recalibration step. The quality control sample (QC) was injected regularly throughout the run after every 4 samples approximately. Data acquisition was settled with an automatic gain control of 5.105 and a C-Trap inject time of 20 ms. The acquisition spectrometric parameters were as follows: the spray voltage (+ 3 kV), the S-Lens RF level (50), the tube lens voltage (+ 100 V), the capillary temperature (350°C), the heater temperature (300°C), the sheath gas pressure (55 arbitrary units), the auxiliary gas flow rate (10 arbitrary units) and the sweep gas flow rate (0 arbitrary units). Full instrument calibration was performed using a MSCAL5 ProteoMassT LTQ/FT-Hybrid ESI Pos/Neg. In addition, Xcalibur V2.2 (Thermo Scientific®, Bremen, Germany) software was used for the generation of all chromatographic peaks acquired in full scan mode. Quality control (QC) samples were used to assess system suitability and monitor instrumental drift throughout the analytical sequence. System performance was evaluated based on QC sample reproducibility, including retention time stability (ΔRT ≤ 0.2 min), mass accuracy (Δm/z ≤ 5 ppm), and signal intensity reproducibility (peak area %CV ≤ 30%). 2.5. Data pre-treatment 2.5.1. NMR MestReNova (v14.1.1) was used for NMR data pretreatment. Non informative zones were removed from NMR spectra such as water region (4.7 to 5 ppm) and maleic acid (5.6 to 6.2 ppm). Alignment step was done to reduce the residual chemical shift effect due to inter-sample pH variations. Spectra were then reduced to integrated regions of equal width (0.02ppm), named “bins”, corresponding to the 0.5 to 9.0 ppm region. 2.5.2. LC-MS Data preprocessing was performed by using workflow4Metabolomics.org (W4M) platform on Galaxy environment. The raw data were at first transformed into a data matrix containing all the peaks present in the samples. CentWave algorithm was used for peak detection, and a “peak grouping” step was done to align the peaks. At this point, undetected ions were integrated with according to m/z and RT through “peak filling” step. Final data matrix table was composed of variables that were repeatable in at least 50% of the samples. Batch correction was done by using Metaboanalyst R package on R environment by using EigenMS as algorithm. At this point, samples were normalized by using dilution factor value calculated through NMR technique and log transformed. 2.6. Metabolites identification 2.6.1. NMR For the NMR platform, the identification of metabolites was carried out using several tools and methods to ensure accuracy. The main tool used was Chenomx Profiler 9.0, allowing to analyze NMR spectra and match them with known metabolites in a database. Additionally, the Human Metabolome Database (HMDB), a free online resource, was used to cross-check and identify metabolites based on their NMR spectra. To further confirm the identification, 2D NMR techniques such as COSY (Correlation Spectroscopy) and HSQC (Heteronuclear Single Quantum Coherence) were employed. These techniques provide detailed information about the molecular structure, helping to validate the metabolites identified using the primary tools. 2.6.2. LC-MS For LC-MS platform metabolites identification MS and MS 2 data were used. In MS data isotopologue and adduct were searched by using CAMERA 21 annotation package on W4M. MS 2 data were generated from pool samples with iterative data dependent MS 2 acquisition (iDDA) and processed through msPurity package 22 included in W4M. All features of interest were searched in MS 2 files with 0.0005 filter for m/z and +/-5s for rt. At this point MS2 spectra was compared to external databases for spectra matching(MassBank https://massbank.eu/MassBank/ , HMDB https://hmdb.ca/ , GNPS http://gnps.ucsd.edu ). Through this process it was possible to reach an annotation level 3 on the Schymanski scale 23 . 2.7. Data normalization Data normalization of urinary samples is a crucial step in ensuring accurate and meaningful interpretation of results. However, determining the most appropriate normalization strategy remains a subject of ongoing debate among scientists. In the context of this study, several methods and algorithms were tested, including the PQN algorithm 24 , creatinine 25 , osmolarity 26 , and dilution factor 27 . The efficiencies of these methods were evaluated through PLS-DA and PLS regression models ( see Supplementary Information Figure S.I.1–2). Based on these results, data matrices obtained from both NMR and LC-MS platforms underwent normalization based on dilution factor values. Dilution factors for individual samples were determined using the NMR technique. For each 1 H-NMR spectrum, the total integral of peaks corresponding to metabolites within the range of 0.5-9.0 ppm was computed. Additionally, the signal of the internal standard (TMSP) at 0 ppm was integrated, and its integral was set to 0. The dilution factor value was computed by calculating the ratio between the integral of all metabolites and the integral of the internal standard. 2.8. Statistical analysis and data integration For both NMR and UHPLC-MS, statistical analysis was conduct by using SIMCA-P software (v17.0, Umetrics, Malmö, Sweden), BioStatFlow webtool (biostatflow.org) and GraphPad Prism version 9.4.1 (GraphPad Software, La Jolla, CA,). Samples included in NMR, MS and data integration analysis are summarized in Fig. 1 . Principal component analysis (PCA) was used to explore samples without any classification knowledge and spot any separation trend, groups, or outliers. PCA score plot was also used to identify “strong outliers” represented by samples placed outside the 0.95 Hotelling’s T 2 ellipse. DmodX was used for the detection of samples exceeding the 0.05 cutoff value defined as critical distance of significance. Orthogonal signal correction (OSC) was applied to discriminant model to remove the inter-subject variability and to describe maximum separation based on class. Its quality was evaluated by the predictability calculated based on the fraction correctly predicted in one-seventh cross-validation (Q²) by considering model with Q²> 0.5 as “good” and Q²> 0.9 as “excellent”. Permutation tests were performed to validate models 28 . For both NMR and MS platforms, OSC-PLS models and their loading plot were used to identify relevant metabolites. Variable important projection (VIP) higher than 1 was considered as significant and considered for univariate statistical analysis. Wilcox-Mann–Whitney U test was performed for comparisons between “progressor” and “stable” groups. Given the exploratory nature of this pilot study and the potential correlation among variables selected through multivariate models, no correction for multiple testing was applied to the p-values of the selected feature. The selected features were worn to generate multivariate receiver operating characteristic (ROC) curves with the aim to evaluate the performance of biomarker models created through automated feature selection. The use of multivariate ROC curve allows to capture the multifaceted nature of disease states and offers a clearer understanding of diagnostic and prognostic capabilities leading to more robust models that better reflect the complexity of diseases and improve predictive accuracy. PLS-DA was used as classification algorithm and univariate T-test as feature ranking method with 2 latent variables. Confusion matrix was used as classifier model for binary response allowing to visually evaluate the performance of classification through the measurement of accuracy, precision and recall values. The use of all these statistical tools and machine learning methods such as OSC-PLS, multivariate ROC curves, and confusion matrices allowed to complement the insights gained from PCA, providing a more robust framework for model development and evaluation. Data integration of NMR and Orbitrap-MS data was investigated through a high-level approach leading to merge results coming from the different blocks 29 , 30 . 3. Results 3.3. Description of the cohort A total of 56 urinary samples collected at 3 months post KTx from 56 KTRs were used. Based on mGFR evolution between 3- and 12-months post KTx, the samples were classified as “progressors” (n = 13) and as “stable” (n = 43) as in Table 1 . Statistical analysis of the KTR cohort, focusing on stable versus progressor status and related clinical data, identified age as the only potential confounding factor. As a result, a multivariate analysis of the entire urinary metabolome was performed, with age used as a discriminant factor, but it revealed no distinct separation between the groups. Table 1 Descriptive statistics on clinical data of Metarein cohort for the 56 patients included in the analysis. For each measurement mean, standard deviation (in bracket) and range are reported. All clinical data refers to samples at 3months visit except for mGFR 12 months which refers to 12 months visit. Clinical measures included: weight (in kg) , height (in cm) , BSA = body surface area, Age (in years), Ethnicity , Sex , mGFR 3months/12 months= GFR measured through plasmatic clearance of 51 Cr-EDTAand deindexed, CreatPenz = enzymatic creatinine, RAC = albumin/creatinine ratio. Total (N = 56) Progressor (N = 13) Stable (N = 43) p-value Age 0.038 Mean (SD) 50.66 (13.57) 43.85 (12.31) 52.72 (13.39) Range 19.00–77.00 29.000–75.000 19.00–77.00 Weight (kg) 0.98 Mean (SD) 74.78 (13.47) 74.86 (13.13) 74.76 (13.73) Range 50.00–108.00 56.80–100.00 50.00–108.00 Height (cm) 0.297 Mean (SD) 170.04 (10.56) 167.35 (10.08) 170.860 (10.67) Range 149.00–193.00 154.00–183.00 149.00–193.00 BSA 0.778 Mean (SD) 1.87 (0.21) 1.86 (0.20) 1.878 (0.213) Range 1.54–2.31 1.63–2.24 1.540–2.310 Ethnicity 0.349 Black 12 (21.4%) 4 (30.8%) 8 (18.6%) White 44 (78.6%) 9 (69.2%) 35 (81.4%) Sex 0.119 Female 20 (35.7%) 7 (53.8%) 13 (30.2%) Male 36 (64.3%) 6 (46.2%) 30 (69.8%) CreatPJaffe 0.367 Mean (SD) 138.75 (40.89) 129.69 (36.05) 141.49 (42.25) Range 69.00–250.00 76.00–222.00 69.00–250.00 AlbuU0 (g/L) 0.565 Mean (SD) 85.98 (114.01) 69.85 (95.37) 90.86 (119.65) Range 5.00–594.00 12.00–305.00 5.00–594.00 mGFR 3m (mL/min) 0.329 Mean (SD) 55.49 (15.87) 59.29 (19.85) 54.34 (14.54) Range 23.30–99.10 33.20–99.10 23.30–81.00 mGFR 12m (mL/min) 0.045 Mean (SD) 55.31 (16.24) 47.44 (15.97) 57.69 (15.73) Range 24.60–89.10 25.60–75.10 24.60 − 89.10 3.2 Metabolomics and multivariate and univariate statistical analysis 3.2.1 NMR To detect outliers, PCA-X, DmodX and Hoteling T 2 models were used. Among the 56 patients, 11 samples were excluded from the multivariate statistical analysis. The high number of outliers was mainly due to intense signals around 1.7 ppm and 2.5 ppm in the NMR spectra, which were significantly higher in these samples compared to the rest of the cohort. These signals were attributed to the metabolite 7-O-mycophenolic acid-β-glucuronide (MPAG) from Mycophenolate Mofetyl, an immunosuppressant used by all patients and 93% excreted in urine. After excluding these outliers, a total of 45 samples were thus included in the multivariate statistical analysis (progressor = 12, stable = 33). A discriminant multivariate statistical analysis was performed by generating an OSC-PLS model (Q 2 = 0.775) ( Fig. 2 a ) which was validated through a permutation test (p-value = 0.0361). A loading plot and a list of VIPs score were generated to select relevant features related to the model (see Supplementary Fig. 1) . Univariate statistical analysis through unpaired Mann-Whitney test was performed on normalized bins by including in the analysis outliers excluded in multivariate data analysis (n = 56). A total of 14 metabolites were assessed as significant for separation between progressor and stable groups. However, among these 14 features only 12 were identified and included in the following models ( Table 2 ). ¹H-NMR metabolite p-value Variation progressor versus stable Serine 0.0001 ↖ Mannitol 0.0002 ↘ 3-methylhistidine 0.0081 ↖ Glycine 0.0117 ↖ Hippurate 0.0157 ↘ 2-aminoadipate 0.0285 ↖ N-phenylacetylglycine 0.0332 ↘ Carnitine 0.0337 ↘ Dimethylamine 0.0367 ↖ Creatinine 0.0386 ↖ TMAO 0.0393 ↘ Choline 0.0426 ↖ Table 2 List of significant features identified through 1 H-NMR platform listed based on p-value . The selected metabolites were used to perform a multivariate receiver operator characteristics (ROC) curve with increasing number of variables to evaluate the performance of our model through automated feature selection ( Fig. 2 b ). The most performant model was reached by using 10 metabolites with an AUC value of 0.794 (95% Cl 0.65–0.929) and a predictive accuracy of 67.8%. Confusion matrix was used to evaluate the performance of the classification model. Using the same set of metabolites employed in multivariate ROC curve, the confusion matrix ( Fig. 2 c) showed 3 ‘progressor’ patients and 17 ‘stable’ patients being misclassified (accuracy = 0.64, precision = 0.37, recall = 0.77). 3.2.2 LC-MS: Positive mode Outlier detection was performed using PCA-X, DmodX, and Hotelling T2 models, and no outliers were identified. As a result, fifty-six samples (n = 56; 13 progressors, 43 stable) were included in multivariate and univariate statistical analyses ( Fig. 1 ) . A supervised OSC-PLS model was run followed by a permutation test (Q2 = 0.933 ; p-value = 0.03073) ( Fig. 3 a ). A VIP score list composed by 753 features was obtained from the discriminant model by using VIP > 1 as threshold. A Wilcoxon-Mann-Whitney test was conduct on VIP by underlining a list of 145 significant features. After annotation, only 36 features were associated to known compounds, and, of these, only 15 were not drugs or redundant metabolites and have been selected as biomarkers. Exogenous metabolites, such as antiviral, analgesic, and antihypertensive drugs, were a minor subset of the non-endogenous features identified as significant VIPs and subsequently excluded from further analysis. As a result, only 15 selected features, which corresponded to identified endogenous metabolites (see Table 3), were used for the subsequent analysis. A multivariate ROC curve was generated (3 variables; AUC = 0.676; 95%Cl 0.416–0.875; predictive accuracy = 53.7%) through automated variable selection ( Fig. 3 b ) . In the confusion matrix ( Fig. 3 c ) , 15 variables were used, yielding an accuracy of 0.71, with 3 “progressor” and 15 “stable” patients misclassified (precision = 0.4, recall = 0.77). Positive mode features p-value Variation progressor versus stable Phenylalanine 0.0027572 ↘︎ Tyrosine (T57) 0.0033591 ↘︎ L-Tryptophan 0.0037193 ↘︎ Kynurenine 0.0072574 ↘︎ Dethiobiotin 0.0075228 ↘︎ Indole 0.01319 ↘︎ Methoxyindoleacetic acid 0.025048 ↘︎ 3-ureidopropionate 0.02628 ↘︎ L-3,4-Dihydroxyphenylalanine methyl ester 0.029779 ↘︎ Mannose 0.033364 ↘︎ L-Valine 0.03476 ↘︎ Hippurate 0.035559 ↘︎ 5-Hydroxyindole-3-acetic acid 0.037831 ↘︎ 2-Hydroxycinnamic acid 0.037939 ↘︎ Methionine 0.04388 ↘︎ Table 3 Significant features annotated for LC-MS analysis in positive mode listed based on p-value. These 15 metabolites were the only non-redundant and endogenous metabolites identified in databases. 3.2.3 LC-MS: Negative mode Fifty-six samples from 56 KTRs were included in multivariate and univariate statistical analyses (n = 56, 13 progressor, 43 stable). According to OSC-PLS (Fig. 4a) discriminant analysis, a good discrimination power between progressor and stable groups was highlighted (Q2 = 0.719). However, the overfitting was evident from the results of the permutation test (p value = 0.5381). As done for the positive mode, a list of 537 VIP (VIP > 1) was obtained. Moreover, a univariate t test using Wilcoxon-Mann-Whitney model was performed on peak intensities. A list of 40 features was found to be significant for discrimination between stable and progressor groups but a major part of them were not identified or corresponding to non-endogenous compounds. Only 4 metabolites were unique features corresponding to biological compounds present in the human body ( Table 4 ). Negative mode features p-value Variation progressor versus stable L-Glutamic acid 0.0077804 ↘ Citrate 0.040444 ↘ N-acetyl-D-tryptophan 0.0154 ↖ Uridine 5'-monophosphate 0.022438 ↘ Table 4 In the table the significant features annotated through LC-MS analysis in negative mode are showed listed based on p-value. Starting from the first 40 features detected as significant in Wilcox-Mann-Whitney t test only these 4 features found a corresponding non-exogenous metabolite in databases. The significant features were used to generate a multivariate ROC curve that show poor model’s performance (4 features, AUC = 0.752, 95% Cl 0.596–0.91, predictive accuracy = 67.7%) compared to the above-detailed data (Fig. 4b). The generated confusion matrix demonstrated, compared to previous models and in line with its ROC curve's result, a lower performance (accuracy = 0.77, precision = 0.5, recall = 0.77) with 3 and 10 patients incorrectly classified (Fig. 4c). 3.4 Data integration A high-level approach was used for data fusion by allowing the integration of results coming from the different blocks. Samples common to NMR and LC-MS platforms were included in the analysis (n = 56, 13 progressor, 43 stable). All the possible combinations of block integration were tested. Finally, the most performant models were obtained by integration of NMR, Positive and Negative MS modes. A total of 31 variables, coming from the 3 different blocks, were integrated, and PCA and OSC-PLS models were performed (see Supplementary Fig. 2 and Supplementary Table 1) . The discrimination between the two groups, that was already visible through the PCA score plot was highlighted by the high discriminant performance of OSC-PLS model (Q2 = 0.829, p-value = 0.00214) ( Fig. 5 a ) . Like previously done for each separate dataset, a multivariate ROC curve was performed ( Fig. 5 b ). The degree of separability was evaluated (20 variables, AUC = 0.816, 95% Cl 0.651–0.947, accuracy = 70.2%) by showing a more performant power compared to models of single blocks. Only 10 “stable” and 2 “progressor” patients were misclassified in confusion matrix by showing a general increased classification power when coupling NMR and LC-MS selected variables (accuracy = 0.81, precision = 0.52, recall = 0.85) ( Fig. 5 c ). By comparing the performance of models from individual platforms and their combinations the integrated model incorporating NMR, both positive and negative modes ( refer to Table 5 ), exhibited the highest efficiency (AUC NMR, POS, NEG = 0.816 versus AUC NMR = 0.794). Specifically, when examining the Q2 values of OSC-PLS models, only the model derived from the Positive mode outperformed the one integrating all 3 platforms. However, despite this, the p-value from the permutation test was lower in the integrated approach, suggesting that the observed result is unlikely to have occurred by random chance alone. Notably, considering the AUC value of the multivariate ROC curve, and the accuracy of the confusion matrix, the integration of NMR and the dual modes LC-MS platform yielded the most robust model. Platform OSC-PLS (Q 2 ) Permutation test (P-value) ROC curve (AUC) CI 95% (AUC) Confusion matrix (accuracy) NMR 0.775 0.03617 0.794 0.650–0.929 0.64 Positive mode (LC-MS) 0.933 0.03073 0.676 0.416–0.875 0.71 Negative mode (LC-MS) 0.719 0.5381 0.752 0.596–0.910 0.77 NMR, LC-MS Positive , LC-MS Negative 0.829 0.00214 0.816 0.651–0.947 0.81 NMR, LC-MS Positive 0.768 0.04432 0.767 0.507–0.907 0.66 NMR, LC-MS Negative 0.783 0.0196 0.812 0.578–0.954 0.70 LC-MS Positive, LC-MS Negative 0.768 0.00618 0.767 0.5-0.919 0.64 Table 5 Comparison of models' performance for integrated and non-integrated models. In the table Q2 value for OSC-PLS models, permutation test p-values a, AUC of multivariate curves, CI 95% of multivariate ROC curves, and accuracy of confusion matrix are reported. All the 31 annotated metabolites were used to generate a pathway analysis to explore the metabolic pathways impacted that could be correlated to the kidney function decrease. In particular, the result of this analysis highlights 11 metabolic pathways be involved in renal function decline with a p-value lower than 0.05 threshold ( Table 6). Pathways Total Hits p-value Aminoacyl-tRNA biosynthesis 48 7 2.69E-07 Phenylalanine metabolism 10 3 0.0001256 Phenylalanine, tyrosine and tryptophan biosynthesis 4 2 0.00067098 Tryptophan metabolism 41 4 0.00078284 Glyoxylate and dicarboxylate metabolism 32 3 0.0044684 Histidine metabolism 16 2 0.012418 Pantothenate and CoA biosynthesis 19 2 0.017357 Glutathione metabolism 28 2 0.036209 Alanine, aspartate and glutamate metabolism 28 2 0.036209 Porphyrin and chlorophyll metabolism 30 2 0.041138 Glycine, serine and threonine metabolism 33 2 0.048981 Table 6 significant metabolic pathways impacted by metabolites coming from the platforms’ integration. ( Total = total number of metabolites composing the pathway, Hits = number of metabolites identified in our model) 4 Discussion This study aims to investigate the potential of metabolomics to predict the decline of measured renal function in KTRs at one year post KTx. We utilized a pilot cohort of well-characterized KTRs with detailed renal function assessments. Access to measured GFR, the gold standard for assessing renal function, clearly represents a key strength of this study. Urine samples were collected 3 months post-transplant, and glomerular filtration rate (GFR) was measured at both 3- and 12-months post-transplant. Based on these mGFR values, patients were categorized into two groups: those with stable renal function and those with 7% reduced mGFR. This 7% number is not arbitrary but based on well-established concepts, like “critical difference” or “least significant change (LSD)”, corresponding to clinically relevant changes in mGFR. To achieve a comprehensive metabolome coverage, we employed a multi-platform analytical strategy. 3 datasets (NMR, MS + and MS-) were thus obtained, analyzed and combined. The multivariate statistical analysis across various platforms allowed us to identify potential biomarkers associated with kidney function decline after transplantation. Notably, the classification of patients using confusion matrices demonstrated high performance and accuracy before metabolite identification. However, many discriminating metabolites used in these models were influenced by external factors, such as medication intake or environmental exposures (e.g., pesticides), rather than renal physiology. Given that patients in this cohort generally follow a similar standard treatment regimen with variations in dosages and comorbidities ( see Supplementary Table 2 ), we subsequently excluded features linked to exogenous compounds, unknown metabolites, and redundant biological substances from the predictive models. At this point, the selected features were used to define predictive and discriminant models for single and integrated datasets. By evidence, the integration of NMR and MS (in positive and negative mode) platforms allowed the model to gain in accuracy for groups prediction. The examination of both the OSC-PLS model and the confusion matrix scores clearly demonstrated the superior discrimination and classification performance of the combined model compared to the single-platform models. Moreover, through the identification of the features on which these models were based, it was possible to delineate a panel of putative biomarkers able to describe and predict kidney function decline from 3 to 12 months post KTx. Even if only Hippurate was found as significant in both platforms, several other common metabolites reached VIP scores. Furthermore, it was shown how metabolites coming from LC-MS and NMR techniques were sharing the same metabolic pathways allowing us to provide complementary information and a global view on metabolome of “progressor” patients. This analysis highlights the fact that all the identified features derive from various metabolic pathways going from amino acids metabolism (serine, glycine, phenylalanine, valine, hippuric acid etc..) to gut microbiota metabolism (TMAO, choline, ) whose importance in CKD frame has been suggested in the last years 31 . The existing link between decline in kidney function and some metabolites highlighted in this study have been demonstrated. For example, TMAO is a well-known metabolite in the context of kidney dysfunction that has been proposed as potential biomarker of CKD in the last decade 32 . Increase in plasmatic TMAO has been reported as strongly linked to CKD 33 , 34 and inversely associated with estimated GFR 35 , 36 . In our study the downregulated TMAO level could be explained by an impaired renal extraction with a consequent depletion of this methylamine in urine, as supported by others 37 , 38 . Similarly, DMA is another methylamine that has been found to be downregulated in the present study. This metabolite has already been described in other works as being correlated to medullary damage and acute rejection post KTx 39 , 40 . Among the metabolites with higher concentrations in the progressor group were choline, glycine, and serine. Elevated choline levels might be linked to decreased TMAO concentration, as choline is metabolized by bacteria to produce TMAO. This increase in choline is consistent with previous studies and may indicate tubulointerstitial dysfunction 41 or accumulation due to increased TMAO formation. Glycine levels might rise from choline generated through phosphatidylcholine metabolism, contributing to its increased circulating concentration and extraction, as demonstrated in earlier research 42 . Higher urinary serine concentrations were also observed in patients with declining kidney function over 12 months. Serine, a precursor in glycine biosynthesis and metabolism, has been previously found at elevated levels in the plasma of CKD patients 42 . Other well-known metabolites in kidney disease are represented by tryptophan and kynurenine. Tryptophane/kynurenine pathways has gained interest in the last years because of its role in acute injury prediction 43 . In addition to this, several other studies demonstrated decrease in tryptophan related to eGFR in serum of patients with kidney dysfunction but no changes in urine biofluids were reported. Other studies reported upregulation of kynurenine or downregulation of serotonin (other tryptophan derived metabolite) according to eGFR impairment in kidney transplanted recipients 44 , 45 . These observations suggest changes in tryptophan/kynurenine pathways associated with kidney GFR impairment. In our study, we hypothesize that the diminution of tryptophan and kynurenine levels in patients with kidney function decline can be linked to major GFR impairment in those patient compared to “stable” one as also showed by Colas et al. 11 who demonstrated an urinary increase in levels of these metabolites in tolerant kidney transplanted recipients compared to non-tolerant ones. Uremic toxin levels in urine decrease with the severity CKD. In this study, we found that 3-methylhistidine and hippuric acid, significant across both platforms, were retained and consequently downregulated in the urine of patients with declining renal function. Hippuric acid, a well-known uremic toxin linked to the gut microbiome, shows reduced urine concentration due to dysregulated active renal tubular secretion 46 . 3-Methylhistidine, associated with muscle protein breakdown, has been implicated in CKD progression and complications, as indicated by increased plasma levels in stage 3–4 CKD patients 47 , . The observed downregulation of these metabolites in "progressor" patients aligns with existing literature, despite limited information on their behavior in urine samples. The primary strength of this study reposed on the high quality of clinical data linked to patients. Notably, to our knowledge, this cohort is among the few documented in the literature where the GFR for each patient is not merely estimated but accurately measured by a reference method. The inclusion of mGFR values for characterizing patients constitutes a significant advantage for a deep and precise understanding of kidney function status. Moreover, the longitudinal nature of the sample cohort has facilitated the establishment of an innovative and clinically relevant patient classification. By employing this novel stratification method, we were able to investigate kidney function evolution in the months following kidney transplantation (KTx) and identify a panel of metabolites that could serve as early predictors of renal function decline in kidney transplant recipients (KTRs). Notably, many of these metabolites have been previously identified or associated with chronic kidney disease (CKD). Another focal point of this study was represented by the platform combination and the data integration that allowed an increase of prediction model performance. Particularly, NMR quantification represented an asset for data normalization constituting a crucial step in urine samples analysis. Indeed, the importance of normalization on creatinine continues to represent an open and fierce debate 27 , 49 , 50 . In the context of this study, several preliminary analyses were conducted before validating the normalization to the dilution factor value. Furthermore, the positive outcomes achieved using the NMR platform convinced us to apply the same normalization procedure to MS data. This approach yielded effective results, thereby validating the robustness of the normalization method. The synergy between these platforms became apparent in our study conclusion, and their application in parallel helped to overcome the primary limitation of NMR represented by its lower sensitivity. Hence, through the combination of NMR's quantitative aspect with MS’s higher sensitivity, a panel of predictive biomarkers elucidating the deterioration of kidney function was established and this will provide a more encompassing vision of its biochemical processes. From a biological perspective, our analysis reveals minimal overlap between platforms, with only hippurate being statistically significant on both. Other metabolites such as creatinine, mannitol, citrate, phenylalanine, tryptophan, and tyrosine appear in the full list of VIPs (data not shown) but did not show significant discrimination in univariate analysis. To illustrate the complementarity of the platforms, a pathway analysis was performed, demonstrating that while different metabolites were identified by each technique, they were all part of the same metabolic pathways. From a clinical point of view, in the stratification process, the “progressor” group is formed based on the mGFR progression, but there is no available information regarding the cause of its decline. Further investigation is warranted to gain a comprehensive understanding of the pathophysiologic events associated with the reduction in GFR within the “progressor” group. Another limitation of this study is the sample size, which poses a significant constraint. While the number of enrolled patients may be deemed reasonable, the uneven distribution of samples across categories complicates the identification of reliable biomarkers for clinical application. Moreover, the single-site collection and the lack of validation cohort limit the robustness of the result, which is critical for routine clinical implementation. Consequently, additional studies and cohorts are mandatory to validate our preliminary findings. From an analytical perspective, 11 out of 56 samples were initially identified as outliers due to the diverse drug regimens, including immunosuppressants. However, since the relevant features for the study were not affected by drug-related regions, all 56 samples were included in the univariate analysis. Despite these limitations, the results generated by this innovative exploratory and pilot study and the metabolites identified as potential biomarkers demonstrate the value of metabolomics and represent a very interesting basis on which to develop more advanced models. 5 Conclusions Early identification of post-transplant renal function trajectories, particularly the ability to predict a decline, is critical for optimizing the management and treatment of transplant recipients. In this context, we hypothesized that metabolomics could serve as a valuable and insightful tool. Consequently, we conducted an exploratory study aimed at predicting adverse renal function outcomes in transplant patients. Despite certain limitations inherent in our methodology, primarily related to the modest cohort size, our findings demonstrate that it is feasible to develop predictive models for renal function decline at 12 months post-transplant by analyzing urine samples collected as early as 3 months post-transplant. On the analytical front, our study highlights the complementary strengths of NMR and MS platforms in metabolomic profiling. Several metabolites identified as potential early biomarkers of renal function degradation have been associated with relevant metabolic pathways implicated in renal pathologies such as CKD. These preliminary findings not only open new avenues for research in KTx but also suggest a potential advancement in clinical diagnostic tools. From a clinical perspective, integrating metabolomics into routine practice could enhance early detection and management of renal function decline. By incorporating predictive models based on early metabolite changes, clinicians could identify at-risk patients earlier and tailor interventions more effectively. This proactive approach may improve patient outcomes by enabling more timely adjustments to treatment regimens and personalized care strategies. Future studies incorporating comprehensive clinical data and validating these models through multi-center or larger cohort studies will be essential to confirm their clinical utility. If successful, these models could become integral to routine post-transplant monitoring, ultimately contributing to better long-term outcomes and more efficient management of kidney transplant patients. Declarations Author information Authors and Affiliations Clinical Metabolomics Group, Center for Interdisciplinary Research on Medicines (CIRM), University of Liege, Liege, Belgium Arianna Cirillo & Pascal de Tullio Division of Nephrology-Dialysis-Transplantation, University of Liège, CHU de Liège, Liège, Belgium Guillaume Resimont, François Jouret & Pierre Delanaye Oniris, INRAE, LABERCA, Nantes, France Justine Massias & Yann Guitton MetaboHUB-MELISA, MetaboHUB-ANR-11-INBS-0010, Oniris, INRAE, LABERCA, Nantes, France Justine Massias & Yann Guitton Interdisciplinary Group for Applied Genoproteomics (GIGA), Cardiovascular Sciences, University of Liège, Liège, Belgium François Jouret Paris Public Hospital System, Renal Physiology Unit, Bichat Hospital Paris, France Emmanuelle Vidal-Petiot & Martin Flamant Paris Cité University and Sorbonne Paris North University, INSERM U1148, LVTS, F-75018 Paris, France Emmanuelle Vidal-Petiot Department of Nephrology-Dialysis-Apheresis, University Hospital Carémeau, Nîmes, France Pierre Delanaye Funding AC and PT are respectively a FRIA grantee and a research director of the Fonds de la Recherche Scientifique-FNRS. FJ received support from the University of Liège (Fonds Spéciaux à la Recherche, Fonds Léon Fredericq) and the FNRS (Research Credits 2016 and 2019). AC received support from the University of Liege (Fonds Léon Fredericq). Contributions PT, FJ, PD, EVP and MF conceived the study. GR, EVP, and MF identified the patients and collected the samples. AC drafted the manuscript. YG and JM performed LC-MS samples and data analysis. AC performed the metabolomics analysis and interpreted the results. PT, FJ, PD, EVP and MF supervised the study. All members of the study group contributed to data collection and provided input into the analysis, writing, and final approval of the manuscript. Corresponding author Correspondence to Pascal de Tullio Ethics and declaration Conflicts of interest All authors state that there is no conflict of interest. Ethical approval For the human cohort, ethical approval for the use of urinary samples and associated metadata in this study was obtained from CEERB Paris Nord (Ethic agreement code: IRB00006477). Data Availability Statement The metabolomics and metadata reported in this paper are available via Metabolight (https://www.ebi.ac.uk/metabolights/) study identifier [REQ20260109215937] References Glassock, R. J., Warnock, D. G., & Delanaye, P. (2017). The global burden of chronic kidney disease: estimates, variability and pitfalls. Nature Reviews Nephrology , 13 , 104–114. Chen, D. Q., et al. (2019). Identification of serum metabolites associating with chronic kidney disease progression and anti-fibrotic effect of 5-methoxytryptophan. Nature Communications , 10 , 1476. Port, F. K., Wolfe, R. A., Mauger, E. A., Berling, D. P., & Jiang, K. (1993). Comparison of survival probabilities for dialysis patients vs cadaveric renal transplant recipients. Journal Of The American Medical Association , 270 , 1339–1343. Horvat, L. D., Shariff, S. Z., & Garg, A. X. (2009). & for the Donor Nephrectomy Outcomes Research (DONOR) Network. Global trends in the rates of living kidney donation. Kidney International , 75 , 1088–1098. Mirzakhani, M., Shahbazi, M., Oliaei, F., & Mohammadnia-Afrouzi, M. (2019). Immunological biomarkers of tolerance in human kidney transplantation: An updated literature review. Journal Cellular Physiology , 234 , 5762–5774. Fiehn, O. C., & Genomics (2001). Metabolome Analysis, and Biochemical Modelling to Understand Metabolic Networks. Comparative and Functional Genomics , 2 , 155–168. Nicholson, J. K., Lindon, J. C., & Holmes, E. (1999). Metabonomics’: understanding the metabolic responses of living systems to pathophysiological stimuli via multivariate statistical analysis of biological NMR spectroscopic data. Xenobiotica , 29 , 1181–1189. Pan, Z., & Raftery, D. (2007). Comparing and combining NMR spectroscopy and mass spectrometry in metabolomics. Analytical And Bioanalytical Chemistry , 387 , 525–527. Riccio, S., et al. (2022). New Insights from Metabolomics in Pediatric Renal Diseases. Children , 9 , 118. Pereira, P. R., et al. (2022). Metabolomics as a tool for the early diagnosis and prognosis of diabetic kidney disease. Medicinal Research Reviews , 42 , 1518–1544. Colas, L., et al. (2022). Urinary metabolomic profiling from spontaneous tolerant kidney transplanted recipients shows enrichment in tryptophan-derived metabolites. eBioMedicine , 77 , 103844. Dahabiyeh, L. A., et al. (2023). Metabolomics profiling distinctively identified end-stage renal disease patients from chronic kidney disease patients. Scientific Reports , 13 , 6161. Kobayashi, T., et al. (2014). A metabolomics-based approach for predicting stages of chronic kidney disease. Biochemical and Biophysical Research Communications , 445 , 412–416. Jouret, F., et al. (2016). Nuclear Magnetic Resonance Metabolomic Profiling of Mouse Kidney, Urine and Serum Following Renal Ischemia/Reperfusion Injury. Plos One , 11 , e0163021. Titan, S. M., et al. (2019). Metabolites related to eGFR: Evaluation of candidate molecules for GFR estimation using untargeted metabolomics. Clinica Chimica Acta , 489 , 242–248. Glavan, M. R. (2025). Targeted Analysis of Serum and Urinary Metabolites for Early Chronic Kidney Disease. IJMS 26, 2862. Meeusen, J. W., Stämmler, F., Dasari, S., Schiffer, E., & Lieske, J. C. (2022). Serum myo-inositol and valine improve metabolomic-based estimated glomerular filtration rate among kidney transplant recipients. Front Med , 9 , 988989. Delanaye, P., Cavalier, E., Depas, G., Chapelle, J. P., & Krzesinski, J. M. (2008). New Data on the Intraindividual Variation of Cystatin C. Nephron. Clinical Practice , 108 , c246–c248. Jiang, L., Huang, J., Wang, Y., & Tang, H. (2012). Eliminating the dication-induced intersample chemical-shift variations for NMR-based biofluid metabonomic analysis. The Analyst , 137 , 4209. Peng, T., Royer, A. L., Guitton, Y., & Le Bizec, B. (2017). Dervilly-Pinel, G. Serum-based metabolomics characterization of pigs treated with ractopamine. Metabolomics , 13 , 77. Kuhl, C., Tautenhahn, R., Böttcher, C., Larson, T. R., & Neumann, S. C. A. M. E. R. A. (2012). An Integrated Strategy for Compound Spectra Extraction and Annotation of Liquid Chromatography/Mass Spectrometry Data Sets. Analytical Chemistry , 84 , 283–289. Lawson, T. N., et al. (2017). msPurity: Automated Evaluation of Precursor Ion Purity for Mass Spectrometry-Based Fragmentation in Metabolomics. Analytical Chemistry , 89 , 2432–2439. Schymanski, E. L., et al. (2014). Identifying Small Molecules via High Resolution Mass Spectrometry: Communicating Confidence. Environmental Science And Technology , 48 , 2097–2098. Dieterle, F., Ross, A., Schlotterbeck, G., & Senn, H. (2006). Probabilistic Quotient Normalization as Robust Method to Account for Dilution of Complex Biological Mixtures. Application in 1 H NMR Metabonomics. Analytical Chemistry , 78 , 4281–4290. Miller, R. C., et al. (2004). Comparison of Specific Gravity and Creatinine for Normalizing Urinary Reproductive Hormone Concentrations. Clinical Chemistry , 50 , 924–932. Comparison of Uncertainties Related to Standardization of Urine Samples with Volume and Creatinine Concentration (2004). The Annals of Occupational Hygiene https://doi.org/10.1093/annhyg/meh019 doi:10.1093/annhyg/meh019. Gagnebin, Y., et al. (2017). Metabolomic analysis of urine samples by UHPLC-QTOF-MS: Impact of normalization strategies. Analytica Chimica Acta , 955 , 27–35. Gavaghan, C. L., Wilson, I. D., & Nicholson, J. K. (2002). Physiological variation in metabolic phenotyping and functional genomic studies: use of orthogonal signal correction and PLS-DA. FEBS Letters , 530 , 191–196. Boccard, J., & Rudaz, S. (2014). Harnessing the complexity of metabolomic data with chemometrics: Metabolomic data analysis with chemometrics. J Chemometrics , 28 , 1–9. Spiteri, M., et al. (2016). Data fusion between high resolution 1H-NMR and mass spectrometry: a synergetic approach to honey botanical origin characterization. Analytical And Bioanalytical Chemistry , 408 , 4389–4401. Hsu, C. N., & Tain, Y. L. (2022). Chronic Kidney Disease and Gut Microbiota: What Is Their Connection in Early Life? IJMS 23, 3954. Zixin, Y., et al. (2022). TMAO as a potential biomarker and therapeutic target for chronic kidney disease: A review. Frontiers In Pharmacology , 13 , 929262. Bain, M. A., Faull, R., Fornasini, G., Milne, R. W., & Evans, A. M. (2006). Accumulation of trimethylamine and trimethylamine-N-oxide in end-stage renal disease patients undergoing haemodialysis. Nephrology Dialysis Transplantation , 21 , 1300–1304. Bell, J. D., et al. (1991). Nuclear magnetic resonance studies of blood plasma and urine from subjects with chronic renal failure: identification of trimethylamine-N-oxide. Biochimica et Biophysica Acta (BBA) - Molecular Basis of Disease , 1096 , 101–107. Stubbs, J. R., et al. (2016). Serum Trimethylamine-N-Oxide is Elevated in CKD and Correlates with Coronary Atherosclerosis Burden. Journal of the American Society of Nephrology , 27 , 305–313. Pelletier, C. C., et al. (2019). Elevation of Trimethylamine-N-Oxide in Chronic Kidney Disease: Contribution of Decreased Glomerular Filtration Rate. Toxins , 11 , 635. Krueger, E. S., Lloyd, T. S., & Tessem, J. S. (2021). The Accumulation and Molecular Effects of Trimethylamine N-Oxide on Metabolic Tissues: It’s Not All Bad. Nutrients , 13 , 2873. Hsu, C. N. (2018). Gut Microbiota-Dependent Trimethylamine N-Oxide Pathway Associated with Cardiovascular Risk in Children with Early-Stage Chronic Kidney Disease. IJMS 19, 3699. Gartland, K. P., Bonner, F. W., & Nicholson, J. K. (1989). Investigations into the biochemical effects of region-specific nephrotoxins. Molecular Pharmacology , 35 , 242–250. Le Moyec, L., et al. (1993). Proton Nuclear Magnetic Resonance Spectroscopy of Urine and Plasma in Renal Transplantation Follow-Up. Nephron , 65 , 433–439. Rhee, E. P., et al. (2013). A Combined Epidemiologic and Metabolomic Approach Improves CKD Prediction. Journal of the American Society of Nephrology , 24 , 1330–1338. Yamaguchi, Y., et al. (2021). Plasma metabolites associated with chronic kidney disease and renal function in adults from the Baltimore Longitudinal Study of Aging. Metabolomics , 17 , 9. Löb, S., & Königsrainer, A. (2009). Role of IDO in Organ Transplantation: Promises and Difficulties. International Reviews of Immunology , 28 , 185–206. Landsberg, A., et al. (2018). Non-invasive staging of chronic kidney allograft damage using urine metabolomic profiling. Pediatric Transplantation , 22 , e13226. Goek, O. N., et al. (2013). Metabolites associate with kidney function decline and incident chronic kidney disease in the general population. Nephrology Dialysis Transplantation , 28 , 2131–2138. Deguchi, T., et al. (2005). Renal Clearance of Endogenous Hippurate Correlates with Expression Levels of Renal Organic Anion Transporters in Uremic Rats. Journal Of Pharmacology And Experimental Therapeutics , 314 , 932–938. Mutsaers, H. A. M., et al. (2013). Optimized Metabolomic Approach to Identify Uremic Solutes in Plasma of Stage 3–4 Chronic Kidney Disease Patients. Plos One , 8 , e71199. Zhang, Z. H., et al. (2016). Metabolomic Signatures of Chronic Kidney Disease of Diverse Etiologies in the Rats and Humans. Journal Of Proteome Research , 15 , 3802–3812. Waikar, S. S., Sabbisetti, V. S., & Bonventre, J. V. (2010). Normalization of urinary biomarkers to creatinine during changes in glomerular filtration rate. Kidney International , 78 , 486–494. Warrack, B. M., et al. (2009). Normalization strategies for metabonomic analysis of urine samples. Journal of Chromatography B , 877 , 547–552. Additional Declarations No competing interests reported. Supplementary Files ACirilloSupplementarymaterial.docx 11306mQACCminimumreportingstandardchecklistNMR.pdf 11306mQACCminimumreportingstandardchecklistLCMS.pdf Cite Share Download PDF Status: Under Revision Version 1 posted Editorial decision: Revision requested 09 Apr, 2026 Reviews received at journal 08 Apr, 2026 Reviews received at journal 03 Apr, 2026 Reviews received at journal 31 Mar, 2026 Reviewers agreed at journal 23 Mar, 2026 Reviewers agreed at journal 23 Mar, 2026 Reviewers agreed at journal 21 Mar, 2026 Reviews received at journal 05 Feb, 2026 Reviewers agreed at journal 26 Jan, 2026 Reviewers agreed at journal 21 Jan, 2026 Reviewers invited by journal 20 Jan, 2026 Editor assigned by journal 16 Jan, 2026 Submission checks completed at journal 16 Jan, 2026 First submitted to journal 16 Jan, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8620894","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":578163676,"identity":"a90ba536-5fbc-4f6c-a868-4cf5de2ef802","order_by":0,"name":"Arianna Cirillo","email":"","orcid":"","institution":"University of Liège","correspondingAuthor":false,"prefix":"","firstName":"Arianna","middleName":"","lastName":"Cirillo","suffix":""},{"id":578163677,"identity":"7a885699-3011-49d2-994d-d943670ed71f","order_by":1,"name":"Guillaume Resimont","email":"","orcid":"","institution":"Centre Hospitalier Universitaire de Liège","correspondingAuthor":false,"prefix":"","firstName":"Guillaume","middleName":"","lastName":"Resimont","suffix":""},{"id":578163678,"identity":"f7225a5b-db30-4dbe-b95f-9ab870225c1a","order_by":2,"name":"Justine Massias","email":"","orcid":"","institution":"Oniris","correspondingAuthor":false,"prefix":"","firstName":"Justine","middleName":"","lastName":"Massias","suffix":""},{"id":578163679,"identity":"ad588c36-aaa8-48eb-b6b8-548437686d7a","order_by":3,"name":"Yann Guitton","email":"","orcid":"","institution":"Oniris","correspondingAuthor":false,"prefix":"","firstName":"Yann","middleName":"","lastName":"Guitton","suffix":""},{"id":578163680,"identity":"7553d68f-cfd3-48af-b2c5-e22da098567f","order_by":4,"name":"François Jouret","email":"","orcid":"","institution":"Centre Hospitalier Universitaire de Liège","correspondingAuthor":false,"prefix":"","firstName":"François","middleName":"","lastName":"Jouret","suffix":""},{"id":578163681,"identity":"5dd0c2b6-3701-45d8-b258-c766c71987ed","order_by":5,"name":"Emmanuelle Vidal-petiot","email":"","orcid":"","institution":"Hôpital Bichat-Claude-Bernard","correspondingAuthor":false,"prefix":"","firstName":"Emmanuelle","middleName":"","lastName":"Vidal-petiot","suffix":""},{"id":578163682,"identity":"fd0a65f6-3074-4cd8-809c-5cceebdb6c78","order_by":6,"name":"Martin Flamant","email":"","orcid":"","institution":"Hôpital Bichat-Claude-Bernard","correspondingAuthor":false,"prefix":"","firstName":"Martin","middleName":"","lastName":"Flamant","suffix":""},{"id":578163683,"identity":"56603018-114e-487e-9d57-ac1024325f71","order_by":7,"name":"Pierre Delanaye","email":"","orcid":"","institution":"Centre Hospitalier Universitaire de Liège","correspondingAuthor":false,"prefix":"","firstName":"Pierre","middleName":"","lastName":"Delanaye","suffix":""},{"id":578163684,"identity":"07ebf992-f9f5-4cd2-9ff2-0f1981739e4f","order_by":8,"name":"Pascal de Tullio","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3klEQVRIiWNgGAWjYHACAyC2kGFjYD7A8ADEZuAhQssBBgkeNga2BIYEkrQAVRowJDAQocWcgXnj449tEjx87D3fJBIKbOz6GXgPPsCnxbKBrdjgIFALG8/ZbRIJBmnJMxv4kg3wuuoAj5kEWItELkjL4WSwCAEt5j8gWnKeAbX8T7YHiRCyhQGqhQ2o5YCdAQNQBK9fmtmKJc6cA/nlmLFFgkFygsRhvmS8DjNnb974oaLMRk6+vfnhjQ9/7Oz523sPfsDrMGY0gcQGdBEMLegC9gQ0jIJRMApGwQgEABVAQF9JB9YJAAAAAElFTkSuQmCC","orcid":"","institution":"University of Liège","correspondingAuthor":true,"prefix":"","firstName":"Pascal","middleName":"","lastName":"de Tullio","suffix":""}],"badges":[],"createdAt":"2026-01-16 16:24:10","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8620894/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8620894/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":100897048,"identity":"d8e4d57f-b575-4b68-807d-ee4383793de7","added_by":"auto","created_at":"2026-01-22 14:21:46","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5590579,"visible":true,"origin":"","legend":"","description":"","filename":"ACirilloManuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/b29f2c81f07e8387a6b6d228.docx"},{"id":100897043,"identity":"5761e2d9-f2f6-4250-b093-a65c9d3b0cc1","added_by":"auto","created_at":"2026-01-22 14:21:46","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10256,"visible":true,"origin":"","legend":"","description":"","filename":"8d44dcdf95d04a0894f2f3ba744dbb94.json","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/628b466bc7b6998628df7bc1.json"},{"id":100949812,"identity":"a63e7bf8-423b-4833-ab28-c28c45c5231b","added_by":"auto","created_at":"2026-01-23 07:05:49","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":202363,"visible":true,"origin":"","legend":"","description":"","filename":"11306mQACCminimumreportingstandardchecklistLCMS.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/c280a8d49cb6b99398941832.pdf"},{"id":101202529,"identity":"4e4c131d-6dcf-4c00-8750-2b32cc5e984b","added_by":"auto","created_at":"2026-01-27 09:35:53","extension":"pdf","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":202411,"visible":true,"origin":"","legend":"","description":"","filename":"11306mQACCminimumreportingstandardchecklistNMR.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/5b361b230100bb7721102ab5.pdf"},{"id":100951301,"identity":"3680e8b9-ad26-4b29-85ae-bd8004d3988e","added_by":"auto","created_at":"2026-01-23 07:10:27","extension":"docx","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1434556,"visible":true,"origin":"","legend":"","description":"","filename":"ACirilloSupplementarymaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/bba059f9891dc57a9d4e6b00.docx"},{"id":100897065,"identity":"0bc9fb6d-a88f-4370-9094-5c5660622d16","added_by":"auto","created_at":"2026-01-22 14:21:46","extension":"xml","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":166242,"visible":true,"origin":"","legend":"","description":"","filename":"8d44dcdf95d04a0894f2f3ba744dbb941enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/de63b166933ce54de9e38a60.xml"},{"id":100897067,"identity":"7143cfbc-3366-462a-9b08-50434bff4c4d","added_by":"auto","created_at":"2026-01-22 14:21:47","extension":"eps","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":239588,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage3.eps","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/601b21d32807c50eda37e2e2.eps"},{"id":100950100,"identity":"18056caf-d3eb-4784-8553-b6e038dabdf2","added_by":"auto","created_at":"2026-01-23 07:06:52","extension":"jpeg","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":420040,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/4ee9e937278fc5de49947447.jpeg"},{"id":100950567,"identity":"0d3fe83d-005e-4077-a5db-1f1912210be4","added_by":"auto","created_at":"2026-01-23 07:08:37","extension":"jpeg","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1074,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/2e1e2db5534c8aaf7e08b13a.jpeg"},{"id":100897046,"identity":"67f4708d-04a6-47d4-a1c3-774f4ca873f4","added_by":"auto","created_at":"2026-01-22 14:21:46","extension":"jpeg","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":58266,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/9e2123265053939447014ce0.jpeg"},{"id":100950176,"identity":"95360e9c-a127-4727-88ec-80039cfde251","added_by":"auto","created_at":"2026-01-23 07:07:07","extension":"jpeg","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":538538,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/dce342b82b9a30360a9a2642.jpeg"},{"id":100897054,"identity":"49a17156-2854-4e96-b00e-fc37fcc6578d","added_by":"auto","created_at":"2026-01-22 14:21:46","extension":"jpeg","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":554689,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/e332c3ef8517c9d0fd4c4931.jpeg"},{"id":100897055,"identity":"b47ade93-1dbc-4ac9-bfd8-3000d16f2c7f","added_by":"auto","created_at":"2026-01-22 14:21:46","extension":"jpeg","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":152278,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/5917f92ff05896ed7444c687.jpeg"},{"id":100897062,"identity":"e994f802-e4df-42c3-a805-d099388597a3","added_by":"auto","created_at":"2026-01-22 14:21:46","extension":"jpeg","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":149948,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/f9f40b621128ad78977807bb.jpeg"},{"id":100951304,"identity":"8f4f7688-36c7-4c5d-bbba-4582386313e9","added_by":"auto","created_at":"2026-01-23 07:10:27","extension":"jpeg","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":41484,"visible":true,"origin":"","legend":"","description":"","filename":"groupimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/1bff28ba2049530939cb6b8e.jpeg"},{"id":100950166,"identity":"89a71f35-70a5-4951-a0d4-cdd17574a0e9","added_by":"auto","created_at":"2026-01-23 07:07:04","extension":"jpeg","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":41860,"visible":true,"origin":"","legend":"","description":"","filename":"groupimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/c8aaf844626f2a15e6a84997.jpeg"},{"id":100950832,"identity":"394a414d-1065-479c-864b-ad0ec24f99bf","added_by":"auto","created_at":"2026-01-23 07:09:21","extension":"jpeg","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":39181,"visible":true,"origin":"","legend":"","description":"","filename":"groupimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/d5eabcb5afecf968fc3db2b6.jpeg"},{"id":100897063,"identity":"a133fc33-634c-4e22-b5de-d86479b001c3","added_by":"auto","created_at":"2026-01-22 14:21:46","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":97257,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/1a8b68755896134d55bab550.png"},{"id":100897057,"identity":"0007591b-14c3-4660-94bf-7b5e495c8c1e","added_by":"auto","created_at":"2026-01-22 14:21:46","extension":"png","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":935,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/9734754e1b541bbcdde67a13.png"},{"id":100897072,"identity":"c0b49c5d-674f-454e-8261-f733180da101","added_by":"auto","created_at":"2026-01-22 14:21:47","extension":"png","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":12420,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/b08a183cc462b63861c9e763.png"},{"id":100897060,"identity":"917846c6-8743-4779-95de-e733c6313324","added_by":"auto","created_at":"2026-01-22 14:21:46","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":112325,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/a73c2bd04a8a9cdaab8b2d2b.png"},{"id":100950880,"identity":"0164595f-70a8-4457-acf1-5758ffba1035","added_by":"auto","created_at":"2026-01-23 07:09:26","extension":"png","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":92101,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/539a7fbf20cea01d6722e714.png"},{"id":100950878,"identity":"6a00ad98-8b7c-4971-901f-398089d458c3","added_by":"auto","created_at":"2026-01-23 07:09:25","extension":"png","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":33316,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/7b17132de5488252df86a4aa.png"},{"id":100897068,"identity":"789f8198-776f-4d06-aab7-7c30c5453ddc","added_by":"auto","created_at":"2026-01-22 14:21:47","extension":"png","order_by":25,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":89862,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/2ae6abf56a62355e5fb44fd7.png"},{"id":100897071,"identity":"519a1d2b-0f78-41f8-9c41-42e3db65c51e","added_by":"auto","created_at":"2026-01-22 14:21:47","extension":"png","order_by":26,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":12296,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinegroupimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/0b2236f6f7224b698cb5a89f.png"},{"id":101207239,"identity":"89560713-1ba5-4a4f-a549-fa561e9da8d0","added_by":"auto","created_at":"2026-01-27 09:58:44","extension":"png","order_by":27,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":11574,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinegroupimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/c9e766f39f2e4e422a516aa6.png"},{"id":100897069,"identity":"3befcf4b-4d5d-4f9d-800f-e8f08cc4d78f","added_by":"auto","created_at":"2026-01-22 14:21:47","extension":"png","order_by":28,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":12084,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinegroupimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/031c5eba9da66009f56412ca.png"},{"id":100897070,"identity":"c0b02575-8e40-4648-adb2-73fe55688010","added_by":"auto","created_at":"2026-01-22 14:21:47","extension":"xml","order_by":29,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":162720,"visible":true,"origin":"","legend":"","description":"","filename":"8d44dcdf95d04a0894f2f3ba744dbb941structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/ad7e30185b925042435e1290.xml"},{"id":100950973,"identity":"d3096733-82ec-49e8-a181-a48d96b18e84","added_by":"auto","created_at":"2026-01-23 07:09:44","extension":"html","order_by":30,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":182587,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/91b1a38e69d40589e72514c8.html"},{"id":100950388,"identity":"e278d235-be6e-4c01-8b3e-6ab7956026d1","added_by":"auto","created_at":"2026-01-23 07:07:55","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":96421,"visible":true,"origin":"","legend":"\u003cp\u003eScheme of samples involved in NMR, LC-MS and integrated models. For NMR analysis only 45 samples were included in multivariate statistical analysis due to the presence of 11 detected outliers. These 11 samples were excluded from multivariate data analysis due to the presence of exogenous molecules hiding spectral zones of interest. The presence of these signals hampers the building of performant multivariate statistical model aiming a correct feature selection. However, after feature selection these samples were again included in univariate statistical analysis and integrated in predictive models.\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/d1f626d6138006f30d8698e0.jpg"},{"id":100897044,"identity":"3f2e012c-f39a-4abb-8b9e-9fb47e4d1506","added_by":"auto","created_at":"2026-01-22 14:21:46","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":126084,"visible":true,"origin":"","legend":"\u003cp\u003eMultivariate statistical analysis of NMR platform. \u003cstrong\u003e(a)\u003c/strong\u003e OSC-PLS score plot of NMR data on decrease and stable groups showed a good performance (Q²=0.775 , \u0026nbsp;p-value=0.03617). (\u003cstrong\u003eb)\u003c/strong\u003e multivariate ROC curve based on the 12 features. \u003cstrong\u003e(c)\u003c/strong\u003e confusion matrix based on the 12 NMR features with accuracy, precision and recall metrics reported.\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/c5c0864298e5290c91737647.jpg"},{"id":100950791,"identity":"e24bfc89-4a7e-4d6a-b5ea-5b6f04ff0642","added_by":"auto","created_at":"2026-01-23 07:09:13","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":120171,"visible":true,"origin":"","legend":"\u003cp\u003eMultivariate statistical analysis of positive MS platform. \u003cstrong\u003e(a) \u003c/strong\u003eOSC-PLS score plot of positive MS data on decrease and stable groups showed a good performance (Q²=0.933 , p-value=0.03073 ).\u003cstrong\u003e(b)\u003c/strong\u003e multivariate ROC curve based on the 15 known annotated metabolites. \u003cstrong\u003e(c) \u003c/strong\u003econfusion matrix based on the positive MS features with accuracy, precision and recall metrics reported.\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/df557c841446e5fdddec5ddc.jpg"},{"id":100897053,"identity":"3d66013c-a565-4626-a488-d269d0e23616","added_by":"auto","created_at":"2026-01-22 14:21:46","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":127945,"visible":true,"origin":"","legend":"\u003cp\u003eMultivariate statistical analysis of negative MS platform. \u003cstrong\u003e(a) \u003c/strong\u003eOSC-PLS score plot of negative MS data on decrease and stable groups showed a good performance (Q²=0.719, p-value=0.5381). \u003cstrong\u003e(b\u003c/strong\u003e) multivariate ROC curve based on the 4 known annotated metabolites. \u003cstrong\u003e(c) \u003c/strong\u003econfusion matrix based on the 4 negative MS features with accuracy, precision and recall metrics reported.\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/935483d8972d5840bb95ee31.jpg"},{"id":100897051,"identity":"9ce731f0-63b7-4ec5-8277-d5179cd41b07","added_by":"auto","created_at":"2026-01-22 14:21:46","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":124685,"visible":true,"origin":"","legend":"\u003cp\u003eMultivariate statistical analysis for integrated platform. (\u003cstrong\u003ea\u003c/strong\u003e) OSC-PLS score plot for integrated platform on decrease and stable groups showed a good performance (Q²=0.829 , p-value=0.00214). \u003cstrong\u003e(b\u003c/strong\u003e) multivariate ROC curve based on the 31 metabolites coming from the three analyses. \u003cstrong\u003e(c) \u003c/strong\u003econfusion matrix based on the 31 positive MS features with accuracy, precision and recall metrics reported.\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/969bc972b5377442a52503a6.jpg"},{"id":101208578,"identity":"f12894af-c559-4123-8ce3-8f75e63f79c5","added_by":"auto","created_at":"2026-01-27 10:10:26","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2416145,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/e050e8f7-7c3b-4283-8002-f24bea881e4a.pdf"},{"id":100950928,"identity":"bfcc9a5f-0531-45d2-bb7b-dee3d2a329eb","added_by":"auto","created_at":"2026-01-23 07:09:37","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":1434556,"visible":true,"origin":"","legend":"","description":"","filename":"ACirilloSupplementarymaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/b13c80549924c0a7cd5cf57c.docx"},{"id":100951019,"identity":"ac1dab9d-e82c-4b84-992e-126be8a8cc3b","added_by":"auto","created_at":"2026-01-23 07:09:50","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":202411,"visible":true,"origin":"","legend":"","description":"","filename":"11306mQACCminimumreportingstandardchecklistNMR.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/066fd6eba06257f94ac8132b.pdf"},{"id":100950076,"identity":"a8f74bdb-c7c5-4834-b95a-04d8bd27b1d0","added_by":"auto","created_at":"2026-01-23 07:06:49","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":202363,"visible":true,"origin":"","legend":"","description":"","filename":"11306mQACCminimumreportingstandardchecklistLCMS.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8620894/v1/c777c29fdaae9d5e0716b0c7.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Multiplatform urine metabolomics for non-invasive prediction of one-year renal function decline in kidney transplant recipients: a pilot study","fulltext":[{"header":"1 Introduction","content":"\u003cp\u003eChronic kidney disease (CKD) is a common disease characterized by chronic alteration and progressive decline in kidney function that affects 5 to 10% of the general population \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. In its most severe form, called end-stage renal disease (ESRD), dialysis or kidney transplantation (KTx) are required \u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. Nowadays, KTx represents the best treatment for patients with ESRD in terms of quality of life, morbidity and mortality \u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Despite advancements, KTx is not devoid of risks, particularly concerning allograft dysfunction, which can result in compromised kidney function and graft rejection. Notably, 30\u0026ndash;40% of patients experience graft loss within 10 years post-KTx\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e,\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. In this regard, monitoring kidney function in kidney transplant recipients (KTRs) is essential and poses a critical concern during the follow-up period. Currently, the absence of biomarkers capable of predicting early kidney function loss post-transplantation is a notable challenge.\u003c/p\u003e \u003cp\u003eIn such a scenario, metabolomics emerges as a promising tool. Metabolomics, as part of the omics sciences, is based on the identification and quantification of small molecules, called \u0026ldquo;metabolites\u0026rdquo; (\u0026lt;\u0026thinsp;1500Da) that describe a biological system at a specific time-point \u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. These metabolites represent the downstream products of a series of biochemical events that occur in an organism under physiological or pathophysiological conditions. Nuclear magnetic resonance (NMR) and mass spectrometry (MS) stand as the primary analytical techniques utilized in metabolomics investigations. NMR platform allows non-targeted, non-destructive, and highly reproducible and robust sample analysis. On the other hand, MS holds a prominent place in metabolomics research due to its heightened sensitivity and its ability to detect and identify a multitude of metabolites within biofluids.\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThanks to advancements in technical instruments and softwares, metabolomics is nowadays increasingly applied to detect disease profiles, including renal dysfunction, with several studies highlighting the link between biomarkers and kidney diseases \u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e,\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Over the past decade, it has become a key tool in understanding CKD, providing insights into disease mechanisms and aiding the identification of diagnostic biomarkers. Many studies have focused on different aspects of CKD, such as disease stage\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e and models for estimating eGFR\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. While CKD has been extensively studied, there is a notable gap in research on kidney function decline in post-KTx, with limited literature available in this area and no predictive model for GFR evolution in KTR. Furthermore, most metabolomics studies performed in the CKD field are based on estimated GFR, which is inherently less precise compared to measured GFR (mGFR), which represents the actual gold standard for the kidney function evaluation.\u003c/p\u003e \u003cp\u003eFor these reasons, we designed this exploratory pilot study to combine the strengths of NMR and MS platforms to analyze urine samples of a well-phenotyped cohort of KTRs, with the expectation of pinpointing early biomarkers that could predict the mid-term (at 1-year post-KTX) deterioration of kidney function; which has been measured and not estimated. For this purpose, we were able to take advantage of 3 months urine samples from KTRs prospectively collected during their protocol visits at 3 and 12 months post-KTx, during which GFR was measured.\u003c/p\u003e"},{"header":"2 Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Patient population and clinical data\u003c/h2\u003e \u003cp\u003eFifty-six KTRs were included from April 2016 to November 2017. Urine samples were collected at 3 months post KTx and GFR was measured at 3 and 12 months after KTx in Bichat Hospital (Paris, France). The study was performed in accordance with relevant guidelines and regulations, and informed consent was obtained from all study participants.\u003c/p\u003e \u003cp\u003ePatients were classified in two groups based on the evolution of measured GFR (mGFR) between 3 and 12 months after KTx. More specifically, mGFR was obtained by measuring the plasma clearance of \u003csup\u003e51\u003c/sup\u003eCrEDTA, which is considered as a reference method. We used the absolute values of mGFR (non-indexed for body surface area) and calculated the variation of GFR between 3 and 12 months (expressed in %). Knowing the relative change of mGFR (ΔGFR), we classified the patients as \u0026ldquo;Progressors\u0026rdquo; or \u0026ldquo;Stable\u0026rdquo;. \u0026ldquo;Progressors\u0026rdquo; were patients whose ΔGFR was declining of more than 7% and other were considered as \u0026ldquo;Stable\u0026rdquo;. The use of 7% as threshold in our stratification method is based on the concept of critical difference obtained at Bichat-Claude-Bernard Hospital were samples and clinical data were collected. The critical difference is based on the intra-individual coefficient of variation of GFR and can be defined as the smallest change in results of GFR which is not due to chance\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Quality control (QC) samples\u003c/h2\u003e \u003cp\u003eQuality control (QC) samples were included to assess analytical reproducibility, batch effects, and system stability for both NMR and LC\u0026ndash;MS analyses.\u003c/p\u003e \u003cdiv id=\"Sec5\" class=\"Section3\"\u003e \u003ch2\u003e2.2.1 NMR\u003c/h2\u003e \u003cp\u003eFor NMR analysis, two complementary QC strategies were employed. A pooled QC sample was generated by combining aliquots from all study participants, providing a representative overview of the cohort\u0026rsquo;s metabolic composition. As sample preparation and acquisition were conducted over two consecutive days, this pooled QC was prepared as two independent pools, each corresponding to one analytical batch.\u003c/p\u003e \u003cp\u003eIn addition, urine samples obtained from individuals external to the study cohort were included as external QCs to monitor instrument performance and inter-day stability independently of cohort-specific biological variability.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section3\"\u003e \u003ch2\u003e2.2.2 LC-MS\u003c/h2\u003e \u003cp\u003e For LC\u0026ndash;MS analysis, QC samples consisted of a pooled urine sample generated from aliquots of all study participants. To further evaluate analytical performance across concentration ranges, diluted pooled QCs (2\u0026times; and 4\u0026times; dilutions) were also prepared and analyzed.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.3 \u003cb\u003eSample preparation\u003c/b\u003e\u003c/h2\u003e \u003cp\u003eMetabolomics were performed on urine samples collected at 3 months in all patients. Urinary samples, corresponding to the second morning urine in fasting conditions. Urines were frozen without additives at \u0026minus;\u0026thinsp;80\u0026deg; C.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section3\"\u003e \u003ch2\u003e2.3.1. NMR\u003c/h2\u003e \u003cp\u003eAliquots of 500\u0026micro;l of urine samples collected at 3-month were thawed on ice, supplemented with KF and prepared by following an optimized procedure for reduce inter-sample chemical-shift variations\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. After this procedure, urine samples were supplemented with 200\u0026micro;l of deuterated phosphate buffer (DPB, pH 7.4), 100\u0026micro;l of a 5mM solution of maleic acid and 10\u0026micro;l of a 10 mg/ml TMSP D\u003csub\u003e2\u003c/sub\u003eO solution for NMR analysis. Urine pH and osmolality were measured before sample analysis. All QC samples were processed using the same protocol as study samples.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section3\"\u003e \u003ch2\u003e2.3.2. LC-MS\u003c/h2\u003e \u003cp\u003eAn aliquot of 500\u0026micro;l of urine sample collected at 3-month was placed in 10kDa centrifugal filter (VWR, Fontenay-sous-Bois, France) and centrifuged at 13000g during 30min at 5\u0026deg;C. Once filtered, internal deuterated standards (leucine-5,5,5-\u003cem\u003ed\u003c/em\u003e3, L-tryptophan 2,3,3 \u003cem\u003ed\u003c/em\u003e3, indole-2,4,5,6,7-\u003cem\u003ed\u003c/em\u003e5-3-acetique acid et 1,14 tetradecanedoic-\u003cem\u003ed\u003c/em\u003e24 acid) were added to each sample and nitrogen blowdown was done. QC samples were prepared and processed in parallel using the same procedure.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e2.4. Sample measurement\u003c/h2\u003e \u003cdiv id=\"Sec11\" class=\"Section3\"\u003e \u003ch2\u003e2.4.1. NMR\u003c/h2\u003e \u003cp\u003eAll samples were recorded at 298 K on a Bruker Avance HD spectrometer operating at 700.17 MHz for the proton signal acquisition. The instrument was equipped with a TCI 5-mm cryoprobe with a Z-gradient. Maleic acid was used as the internal standard for quantification and trimethylsilyl-3-propionic acid-\u003cem\u003ed\u003c/em\u003e4 (TMSP) for the zero for the zero calibration. \u003csup\u003e1\u003c/sup\u003eH-NMR spectra were acquired using a 1D NOESY sequence with presaturation. The Noesypresat experiment used a RD-90\u0026deg;-T1-90\u0026deg;-Tm-90\u0026deg;-acquire sequence with a relaxation delay of 4 s, a mixing time (Tm) of 10 ms and a fixed T1 delay of 4 \u0026micro;s. Water suppression pulse was placed during the relaxation delay (RD). The number of transients is 64 (64K data points). QC samples were injected at regular intervals, every 10 samples, throughout the analytical sequence. The data were processed with the Bruker Topspin 4.0.8 software with a standard parameter set. Phase and baseline corrections were performed manually over the entire range of the spectra and the δ scale was calibrated to 0 ppm using the internal standard TMSP. System suitability and inter-batch instrumental drift was assessed by using QC samples. System performance was verified using the TMSP signal at 0.00 ppm based on chemical shift stability (\u0026plusmn;\u0026thinsp;0.01 ppm), line width (FWHM\u0026thinsp;\u0026le;\u0026thinsp;1.2 Hz), and integral reproducibility (%CV\u0026thinsp;\u0026le;\u0026thinsp;10%).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section3\"\u003e \u003ch2\u003e2.4.2. LC-MS\u003c/h2\u003e \u003cp\u003eAll samples were analyzed on ultrahigh performance liquid chromatography with high-resolution mass spectrometry (UHLPC/MS) by following described method\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e for reversed phase(RP) UHPLC/MS. Tacking advantage of the MS2 capacities of the hybrid\u003c/p\u003e \u003cp\u003equadrupole-orbitrap (Q-Exactive TM) mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) QC samples (i.e.. pooled samples) were analyzed, in ESI positive and ESI negative modes, with three cycles of iterative Data Dependent MS\u003csup\u003e2\u003c/sup\u003e. The acquisition of the raw data was performed using a full scan mode within the m/z 65-1000 range at a resolving power of 70,000 at m/z 200,000 Da. Compound separation was performed using a Hypersil GOLD-C18 column (1.9 \u0026micro;m, 100 mm x 2.1 mm) from Thermo-Scientific (USA). The column temperature was set at 35\u0026deg;C. The mobile phases were composed of 0.1% of acetic acid in water (solvent A) and in acetonitrile (solvent B). Acetonitrile LC/HRMS grade CHROMASOLV\u0026trade; LC-MS (Riedel-de Ha\u0026euml;n), Water LC/HRMS grade CHROMANORM\u0026reg; (VWR Chemicals). The applied gradient (A:B, v/v) was as follows: 95:5 from 0 to 2.4 min, 75:25 at 4.5 min, 25:75 at 11 min, 0:100 from 14 at 16.5 min and 95:5 from 19 to 25 min. The flow rate was set to 0.40 mL/min. The injection volume was 5 \u0026micro;L. All samples were analyzed in one batch without any stopping or recalibration step. The quality control sample (QC) was injected regularly throughout the run after every 4 samples approximately. Data acquisition was settled with an automatic gain control of 5.105 and a C-Trap inject time of 20 ms. The acquisition spectrometric parameters were as follows: the spray voltage (+\u0026thinsp;3 kV), the S-Lens RF level (50), the tube lens voltage (+\u0026thinsp;100 V), the capillary temperature (350\u0026deg;C), the heater temperature (300\u0026deg;C), the sheath gas pressure (55 arbitrary units), the auxiliary gas flow rate (10 arbitrary units) and the sweep gas flow rate (0 arbitrary units). Full instrument calibration was performed using a MSCAL5 ProteoMassT LTQ/FT-Hybrid ESI Pos/Neg. In addition, Xcalibur V2.2 (Thermo Scientific\u0026reg;, Bremen, Germany) software was used for the generation of all chromatographic peaks acquired in full scan mode. Quality control (QC) samples were used to assess system suitability and monitor instrumental drift throughout the analytical sequence. System performance was evaluated based on QC sample reproducibility, including retention time stability (ΔRT\u0026thinsp;\u0026le;\u0026thinsp;0.2 min), mass accuracy (Δm/z\u0026thinsp;\u0026le;\u0026thinsp;5 ppm), and signal intensity reproducibility (peak area %CV\u0026thinsp;\u0026le;\u0026thinsp;30%).\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e2.5. Data pre-treatment\u003c/h2\u003e \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e \u003ch2\u003e2.5.1. NMR\u003c/h2\u003e \u003cp\u003eMestReNova (v14.1.1) was used for NMR data pretreatment. Non informative zones were removed from NMR spectra such as water region (4.7 to 5 ppm) and maleic acid (5.6 to 6.2 ppm). Alignment step was done to reduce the residual chemical shift effect due to inter-sample pH variations. Spectra were then reduced to integrated regions of equal width (0.02ppm), named \u0026ldquo;bins\u0026rdquo;, corresponding to the 0.5 to 9.0 ppm region.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section3\"\u003e \u003ch2\u003e2.5.2. LC-MS\u003c/h2\u003e \u003cp\u003eData preprocessing was performed by using workflow4Metabolomics.org (W4M) platform on Galaxy environment. The raw data were at first transformed into a data matrix containing all the peaks present in the samples. CentWave algorithm was used for peak detection, and a \u0026ldquo;peak grouping\u0026rdquo; step was done to align the peaks. At this point, undetected ions were integrated with according to m/z and RT through \u0026ldquo;peak filling\u0026rdquo; step. Final data matrix table was composed of variables that were repeatable in at least 50% of the samples. Batch correction was done by using Metaboanalyst R package on R environment by using EigenMS as algorithm.\u003c/p\u003e \u003cp\u003eAt this point, samples were normalized by using dilution factor value calculated through NMR technique and log transformed.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e2.6. Metabolites identification\u003c/h2\u003e \u003cdiv id=\"Sec17\" class=\"Section3\"\u003e \u003ch2\u003e2.6.1. NMR\u003c/h2\u003e \u003cp\u003eFor the NMR platform, the identification of metabolites was carried out using several tools and methods to ensure accuracy. The main tool used was Chenomx Profiler 9.0, allowing to analyze NMR spectra and match them with known metabolites in a database. Additionally, the Human Metabolome Database (HMDB), a free online resource, was used to cross-check and identify metabolites based on their NMR spectra. To further confirm the identification, 2D NMR techniques such as COSY (Correlation Spectroscopy) and HSQC (Heteronuclear Single Quantum Coherence) were employed. These techniques provide detailed information about the molecular structure, helping to validate the metabolites identified using the primary tools.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section3\"\u003e \u003ch2\u003e2.6.2. LC-MS\u003c/h2\u003e \u003cp\u003eFor LC-MS platform metabolites identification MS and MS\u003csup\u003e2\u003c/sup\u003e data were used. In MS data isotopologue and adduct were searched by using CAMERA\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e annotation package on W4M. MS\u003csup\u003e2\u003c/sup\u003e data were generated from pool samples with iterative data dependent MS\u003csup\u003e2\u003c/sup\u003e acquisition (iDDA) and processed through msPurity package\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e included in W4M. All features of interest were searched in MS\u003csup\u003e2\u003c/sup\u003e files with 0.0005 filter for m/z and +/-5s for rt. At this point MS2 spectra was compared to external databases for spectra matching(MassBank \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://massbank.eu/MassBank/\u003c/span\u003e\u003cspan address=\"https://massbank.eu/MassBank/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e, HMDB \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://hmdb.ca/\u003c/span\u003e\u003cspan address=\"https://hmdb.ca/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e, GNPS \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://gnps.ucsd.edu\u003c/span\u003e\u003cspan address=\"http://gnps.ucsd.edu\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). \u003cem\u003eThrough this process it was possible to reach an annotation level 3 on the\u003c/em\u003e Schymanski scale\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003e2.7. Data normalization\u003c/h2\u003e \u003cp\u003eData normalization of urinary samples is a crucial step in ensuring accurate and meaningful interpretation of results. However, determining the most appropriate normalization strategy remains a subject of ongoing debate among scientists. In the context of this study, several methods and algorithms were tested, including the PQN algorithm\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e, creatinine\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e, osmolarity\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e, and dilution factor\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. The efficiencies of these methods were evaluated through PLS-DA and PLS regression models (\u003cb\u003esee Supplementary Information Figure S.I.1\u0026ndash;2).\u003c/b\u003e Based on these results, data matrices obtained from both NMR and LC-MS platforms underwent normalization based on dilution factor values. Dilution factors for individual samples were determined using the NMR technique. For each \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003eH-NMR spectrum, the total integral of peaks corresponding to metabolites within the range of 0.5-9.0 ppm was computed. Additionally, the signal of the internal standard (TMSP) at 0 ppm was integrated, and its integral was set to 0. The dilution factor value was computed by calculating the ratio between the integral of all metabolites and the integral of the internal standard.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003e2.8. Statistical analysis and data integration\u003c/h2\u003e \u003cp\u003eFor both NMR and UHPLC-MS, statistical analysis was conduct by using SIMCA-P software (v17.0, Umetrics, Malm\u0026ouml;, Sweden), BioStatFlow webtool (biostatflow.org) and GraphPad Prism version 9.4.1 (GraphPad Software, La Jolla, CA,). Samples included in NMR, MS and data integration analysis are summarized in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Principal component analysis (PCA) was used to explore samples without any classification knowledge and spot any separation trend, groups, or outliers. PCA score plot was also used to identify \u0026ldquo;strong outliers\u0026rdquo; represented by samples placed outside the 0.95 Hotelling\u0026rsquo;s T\u003csup\u003e2\u003c/sup\u003e ellipse. DmodX was used for the detection of samples exceeding the 0.05 cutoff value defined as critical distance of significance.\u003c/p\u003e \u003cp\u003eOrthogonal signal correction (OSC) was applied to discriminant model to remove the inter-subject variability and to describe maximum separation based on class. Its quality was evaluated by the predictability calculated based on the fraction correctly predicted in one-seventh cross-validation (Q\u0026sup2;) by considering model with Q\u0026sup2;\u0026gt; 0.5 as \u0026ldquo;good\u0026rdquo; and Q\u0026sup2;\u0026gt; 0.9 as \u0026ldquo;excellent\u0026rdquo;. Permutation tests were performed to validate models \u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. For both NMR and MS platforms, OSC-PLS models and their loading plot were used to identify relevant metabolites. Variable important projection (VIP) higher than 1 was considered as significant and considered for univariate statistical analysis. Wilcox-Mann\u0026ndash;Whitney U test was performed for comparisons between \u0026ldquo;progressor\u0026rdquo; and \u0026ldquo;stable\u0026rdquo; groups. Given the exploratory nature of this pilot study and the potential correlation among variables selected through multivariate models, no correction for multiple testing was applied to the p-values of the selected feature. The selected features were worn to generate multivariate receiver operating characteristic (ROC) curves with the aim to evaluate the performance of biomarker models created through automated feature selection. The use of multivariate ROC curve allows to capture the multifaceted nature of disease states and offers a clearer understanding of diagnostic and prognostic capabilities leading to more robust models that better reflect the complexity of diseases and improve predictive accuracy.\u003c/p\u003e \u003cp\u003ePLS-DA was used as classification algorithm and univariate T-test as feature ranking method with 2 latent variables. Confusion matrix was used as classifier model for binary response allowing to visually evaluate the performance of classification through the measurement of accuracy, precision and recall values. The use of all these statistical tools and machine learning methods such as OSC-PLS, multivariate ROC curves, and confusion matrices allowed to complement the insights gained from PCA, providing a more robust framework for model development and evaluation. Data integration of NMR and Orbitrap-MS data was investigated through a high-level approach leading to merge results coming from the different blocks \u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e,\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e3.3. Description of the cohort\u003c/h2\u003e \u003cp\u003eA total of 56 urinary samples collected at 3 months post KTx from 56 KTRs were used. Based on mGFR evolution between 3- and 12-months post KTx, the samples were classified as \u0026ldquo;progressors\u0026rdquo; (n\u0026thinsp;=\u0026thinsp;13) and as \u0026ldquo;stable\u0026rdquo; (n\u0026thinsp;=\u0026thinsp;43) as in \u003cb\u003eTable\u0026nbsp;1\u003c/b\u003e. Statistical analysis of the KTR cohort, focusing on stable \u003cem\u003eversus\u003c/em\u003e progressor status and related clinical data, identified age as the only potential confounding factor. As a result, a multivariate analysis of the entire urinary metabolome was performed, with age used as a discriminant factor, but it revealed no distinct separation between the groups.\u003c/p\u003e \u003cp\u003e\u003cstrong\u003eTable\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e1\u003c/strong\u003e Descriptive statistics on clinical data of Metarein cohort for the 56 patients included in the analysis. For each measurement mean, standard deviation (in bracket) and range are reported. All clinical data refers to samples at 3months visit except for\u0026nbsp;\u003cstrong\u003emGFR 12 months\u0026nbsp;\u003c/strong\u003ewhich refers to 12 months visit. Clinical measures included: \u003cstrong\u003eweight\u003c/strong\u003e (in kg) , \u003cstrong\u003eheight\u003c/strong\u003e (in cm) , \u003cstrong\u003eBSA\u003c/strong\u003e= body surface area, \u003cstrong\u003eAge\u003c/strong\u003e (in years), \u003cstrong\u003eEthnicity\u003c/strong\u003e,\u003cstrong\u003e\u0026nbsp;Sex\u003c/strong\u003e,\u0026nbsp;\u003cstrong\u003emGFR\u003csub\u003e\u0026nbsp;\u003c/sub\u003e3months/12 months=\u0026nbsp;\u003c/strong\u003eGFR measured through plasmatic clearance of \u003csup\u003e51\u003c/sup\u003eCr-EDTAand deindexed, \u003cstrong\u003e\u0026nbsp;CreatPenz\u003c/strong\u003e= enzymatic creatinine, \u003cstrong\u003eRAC\u003c/strong\u003e= albumin/creatinine ratio.\u003c/p\u003e\u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Taba\" border=\"1\"\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTotal (N\u0026thinsp;=\u0026thinsp;56)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eProgressor (N\u0026thinsp;=\u0026thinsp;13)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eStable (N\u0026thinsp;=\u0026thinsp;43)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003ep-value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAge\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.038\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e50.66 (13.57)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e43.85 (12.31)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e52.72 (13.39)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRange\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e19.00\u0026ndash;77.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e29.000\u0026ndash;75.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e19.00\u0026ndash;77.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eWeight (kg)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.98\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e74.78 (13.47)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e74.86 (13.13)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e74.76 (13.73)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRange\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e50.00\u0026ndash;108.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e56.80\u0026ndash;100.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e50.00\u0026ndash;108.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eHeight (cm)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.297\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e170.04 (10.56)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e167.35 (10.08)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e170.860 (10.67)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRange\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e149.00\u0026ndash;193.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e154.00\u0026ndash;183.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e149.00\u0026ndash;193.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eBSA\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.778\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.87 (0.21)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.86 (0.20)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.878 (0.213)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRange\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.54\u0026ndash;2.31\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.63\u0026ndash;2.24\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.540\u0026ndash;2.310\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eEthnicity\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.349\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBlack\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e12 (21.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4 (30.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e8 (18.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWhite\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e44 (78.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e9 (69.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e35 (81.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eSex\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.119\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e20 (35.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e7 (53.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e13 (30.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e36 (64.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e6 (46.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e30 (69.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCreatPJaffe\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.367\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e138.75 (40.89)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e129.69 (36.05)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e141.49 (42.25)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRange\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e69.00\u0026ndash;250.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e76.00\u0026ndash;222.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e69.00\u0026ndash;250.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAlbuU0 (g/L)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.565\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e85.98 (114.01)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e69.85 (95.37)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e90.86 (119.65)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRange\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5.00\u0026ndash;594.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e12.00\u0026ndash;305.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e5.00\u0026ndash;594.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003emGFR 3m (mL/min)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.329\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e55.49 (15.87)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e59.29 (19.85)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e54.34 (14.54)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRange\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e23.30\u0026ndash;99.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e33.20\u0026ndash;99.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e23.30\u0026ndash;81.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003emGFR 12m\u003c/b\u003e \u003cb\u003e(mL/min)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.045\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e55.31 (16.24)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e47.44 (15.97)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e57.69 (15.73)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRange\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e24.60\u0026ndash;89.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e25.60\u0026ndash;75.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e24.60 \u0026minus;\u0026thinsp;89.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec23\" class=\"Section2\"\u003e \u003ch2\u003e\u003cb\u003e3.2 Metabolomics and multivariate and univariate statistical analysis\u003c/b\u003e\u003c/h2\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec24\" class=\"Section3\"\u003e \u003ch2\u003e3.2.1 NMR\u003c/h2\u003e \u003cp\u003eTo detect outliers, PCA-X, DmodX and Hoteling T\u003csup\u003e2\u003c/sup\u003e models were used. Among the 56 patients, 11 samples were excluded from the multivariate statistical analysis. The high number of outliers was mainly due to intense signals around 1.7 ppm and 2.5 ppm in the NMR spectra, which were significantly higher in these samples compared to the rest of the cohort. These signals were attributed to the metabolite 7-O-mycophenolic acid-β-glucuronide (MPAG) from Mycophenolate Mofetyl, an immunosuppressant used by all patients and 93% excreted in urine. After excluding these outliers, a total of 45 samples were thus included in the multivariate statistical analysis (progressor\u0026thinsp;=\u0026thinsp;12, stable\u0026thinsp;=\u0026thinsp;33).\u003c/p\u003e \u003cp\u003eA discriminant multivariate statistical analysis was performed by generating an OSC-PLS model (Q\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.775) \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea\u003cb\u003e)\u003c/b\u003e which was validated through a permutation test (p-value\u0026thinsp;=\u0026thinsp;0.0361). A loading plot and a list of VIPs score were generated to select relevant features related to the model \u003cb\u003e(see Supplementary Fig.\u0026nbsp;1)\u003c/b\u003e. Univariate statistical analysis through unpaired Mann-Whitney test was performed on normalized bins by including in the analysis outliers excluded in multivariate data analysis (n\u0026thinsp;=\u0026thinsp;56). A total of 14 metabolites were assessed as significant for separation between progressor and stable groups. However, among these 14 features only 12 were identified and included in the following models (\u003cb\u003eTable\u0026nbsp;2\u003c/b\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabb\" border=\"1\"\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026sup1;H-NMR metabolite\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ep-value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eVariation\u003c/p\u003e \u003cp\u003eprogressor versus stable\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSerine\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↖\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMannitol\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0002\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↘\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3-methylhistidine\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0081\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↖\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGlycine\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0117\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↖\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHippurate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0157\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↘\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2-aminoadipate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0285\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↖\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eN-phenylacetylglycine\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0332\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↘\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCarnitine\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0337\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↘\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDimethylamine\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0367\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↖\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCreatinine\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0386\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↖\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTMAO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0393\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↘\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCholine\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0426\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↖\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e\u003cstrong\u003eTable\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e2\u003c/strong\u003e\u0026nbsp; List of significant features identified through \u003csup\u003e1\u003c/sup\u003eH-NMR platform listed based on p-value .\u003c/p\u003e\u003cp\u003eThe selected metabolites were used to perform a multivariate receiver operator characteristics (ROC) curve with increasing number of variables to evaluate the performance of our model through automated feature selection \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb\u003cb\u003e).\u003c/b\u003e The most performant model was reached by using 10 metabolites with an AUC value of 0.794 (95% Cl 0.65\u0026ndash;0.929) and a predictive accuracy of 67.8%. Confusion matrix was used to evaluate the performance of the classification model. Using the same set of metabolites employed in multivariate ROC curve, the confusion\u003c/p\u003e \u003cp\u003ematrix \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec) showed 3 \u0026lsquo;progressor\u0026rsquo; patients and 17 \u0026lsquo;stable\u0026rsquo; patients being misclassified (accuracy\u0026thinsp;=\u0026thinsp;0.64, precision\u0026thinsp;=\u0026thinsp;0.37, recall\u0026thinsp;=\u0026thinsp;0.77).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e \u003ch2\u003e3.2.2 LC-MS: Positive mode\u003c/h2\u003e \u003cp\u003eOutlier detection was performed using PCA-X, DmodX, and Hotelling T2 models, and no outliers were identified. As a result, fifty-six samples (n\u0026thinsp;=\u0026thinsp;56; 13 progressors, 43 stable) were included in multivariate and univariate statistical analyses \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e\u003cb\u003e)\u003c/b\u003e. A supervised OSC-PLS model was run followed by a permutation test (Q2\u0026thinsp;=\u0026thinsp;0.933 ; p-value\u0026thinsp;=\u0026thinsp;0.03073) \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea\u003cb\u003e).\u003c/b\u003e A VIP score list composed by 753 features was obtained from the discriminant model by using VIP\u0026thinsp;\u0026gt;\u0026thinsp;1 as threshold. A Wilcoxon-Mann-Whitney test was conduct on VIP by underlining a list of 145 significant features.\u003c/p\u003e \u003cp\u003eAfter annotation, only 36 features were associated to known compounds, and, of these, only 15 were not drugs or redundant metabolites and have been selected as biomarkers. Exogenous metabolites, such as antiviral, analgesic, and antihypertensive drugs, were a minor subset of the non-endogenous features identified as significant VIPs and subsequently excluded from further analysis. As a result, only 15 selected features, which corresponded to identified endogenous metabolites (see Table\u0026nbsp;3), were used for the subsequent analysis. A multivariate ROC curve was generated (3 variables; AUC\u0026thinsp;=\u0026thinsp;0.676; 95%Cl 0.416\u0026ndash;0.875; predictive accuracy\u0026thinsp;=\u0026thinsp;53.7%) through automated variable selection \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb\u003cb\u003e)\u003c/b\u003e. In the confusion matrix \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec\u003cb\u003e)\u003c/b\u003e, 15 variables were used, yielding an accuracy of 0.71, with 3 \u0026ldquo;progressor\u0026rdquo; and 15 \u0026ldquo;stable\u0026rdquo; patients misclassified (precision\u0026thinsp;=\u0026thinsp;0.4, recall\u0026thinsp;=\u0026thinsp;0.77).\u003c/p\u003e \u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" align=\"\" width=\"675\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 310px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePositive mode features\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 173px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ep-value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 191px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVariation\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eprogressor versus stable\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 310px;\"\u003e\n \u003cp\u003ePhenylalanine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 173px;\"\u003e\n \u003cp\u003e0.0027572\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 191px;\"\u003e\n \u003cp\u003e↘︎\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 310px;\"\u003e\n \u003cp\u003eTyrosine (T57)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 173px;\"\u003e\n \u003cp\u003e0.0033591\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 191px;\"\u003e\n \u003cp\u003e↘︎\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 310px;\"\u003e\n \u003cp\u003eL-Tryptophan\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 173px;\"\u003e\n \u003cp\u003e0.0037193\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 191px;\"\u003e\n \u003cp\u003e↘︎\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 310px;\"\u003e\n \u003cp\u003eKynurenine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 173px;\"\u003e\n \u003cp\u003e0.0072574\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 191px;\"\u003e\n \u003cp\u003e↘︎\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 310px;\"\u003e\n \u003cp\u003eDethiobiotin\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 173px;\"\u003e\n \u003cp\u003e0.0075228\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 191px;\"\u003e\n \u003cp\u003e↘︎\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 310px;\"\u003e\n \u003cp\u003eIndole\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 173px;\"\u003e\n \u003cp\u003e0.01319\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 191px;\"\u003e\n \u003cp\u003e↘︎\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 310px;\"\u003e\n \u003cp\u003eMethoxyindoleacetic acid\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 173px;\"\u003e\n \u003cp\u003e0.025048\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 191px;\"\u003e\n \u003cp\u003e↘︎\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 310px;\"\u003e\n \u003cp\u003e3-ureidopropionate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 173px;\"\u003e\n \u003cp\u003e0.02628\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 191px;\"\u003e\n \u003cp\u003e↘︎\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 310px;\"\u003e\n \u003cp\u003eL-3,4-Dihydroxyphenylalanine methyl ester\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 173px;\"\u003e\n \u003cp\u003e0.029779\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 191px;\"\u003e\n \u003cp\u003e↘︎\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 310px;\"\u003e\n \u003cp\u003eMannose\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 173px;\"\u003e\n \u003cp\u003e0.033364\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 191px;\"\u003e\n \u003cp\u003e↘︎\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 310px;\"\u003e\n \u003cp\u003eL-Valine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 173px;\"\u003e\n \u003cp\u003e0.03476\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 191px;\"\u003e\n \u003cp\u003e↘︎\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 310px;\"\u003e\n \u003cp\u003eHippurate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 173px;\"\u003e\n \u003cp\u003e0.035559\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 191px;\"\u003e\n \u003cp\u003e↘︎\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 310px;\"\u003e\n \u003cp\u003e5-Hydroxyindole-3-acetic acid\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 173px;\"\u003e\n \u003cp\u003e0.037831\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 191px;\"\u003e\n \u003cp\u003e↘︎\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 310px;\"\u003e\n \u003cp\u003e2-Hydroxycinnamic acid\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 173px;\"\u003e\n \u003cp\u003e0.037939\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 191px;\"\u003e\n \u003cp\u003e↘︎\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 310px;\"\u003e\n \u003cp\u003eMethionine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 173px;\"\u003e\n \u003cp\u003e0.04388\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 191px;\"\u003e\n \u003cp\u003e↘︎\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eTable\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e3\u003c/strong\u003e Significant features annotated for LC-MS analysis in positive mode listed based on p-value. These 15 metabolites were the only non-redundant and endogenous metabolites identified in databases.\u003c/p\u003e\u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e \u003ch2\u003e3.2.3 LC-MS: Negative mode\u003c/h2\u003e \u003cp\u003eFifty-six samples from 56 KTRs were included in multivariate and univariate statistical analyses (n\u0026thinsp;=\u0026thinsp;56, 13 progressor, 43 stable). According to OSC-PLS \u003cb\u003e(Fig.\u0026nbsp;4a)\u003c/b\u003e discriminant analysis, a good discrimination power between progressor and stable groups was highlighted (Q2\u0026thinsp;=\u0026thinsp;0.719). However, the overfitting was evident from the results of the permutation test (p value\u0026thinsp;=\u0026thinsp;0.5381). As done for the positive mode, a list of 537 VIP (VIP\u0026thinsp;\u0026gt;\u0026thinsp;1) was obtained. Moreover, a univariate t test using Wilcoxon-Mann-Whitney model was performed on peak intensities. A list of 40 features was found to be significant for discrimination between stable and progressor groups but a major part of them were not identified or corresponding to non-endogenous compounds. Only 4 metabolites were unique features corresponding to biological compounds present in the human body (\u003cb\u003eTable\u0026nbsp;4\u003c/b\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabc\" border=\"1\"\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNegative mode features\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ep-value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eVariation\u003c/p\u003e \u003cp\u003eprogressor \u003cem\u003eversus\u003c/em\u003e stable\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eL-Glutamic acid\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0077804\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↘\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCitrate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.040444\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↘\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eN-acetyl-D-tryptophan\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.0154\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↖\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUridine 5'-monophosphate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.022438\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e↘\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTable\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e4\u003c/strong\u003e In the table the significant features annotated through LC-MS analysis in negative mode are showed listed based on p-value. Starting from the first 40 features detected as significant in Wilcox-Mann-Whitney t test only these 4 features found a corresponding non-exogenous metabolite in databases.\u003c/p\u003e \u003cp\u003eThe significant features were used to generate a multivariate ROC curve that show poor model\u0026rsquo;s performance (4 features, AUC\u0026thinsp;=\u0026thinsp;0.752, 95% Cl 0.596\u0026ndash;0.91, predictive accuracy\u0026thinsp;=\u0026thinsp;67.7%) compared to the above-detailed data \u003cb\u003e(Fig.\u0026nbsp;4b).\u003c/b\u003e The generated confusion matrix demonstrated, compared to previous models and in line with its ROC curve's result, a lower performance (accuracy\u0026thinsp;=\u0026thinsp;0.77, precision\u0026thinsp;=\u0026thinsp;0.5, recall\u0026thinsp;=\u0026thinsp;0.77) with 3 and 10 patients incorrectly classified \u003cb\u003e(Fig.\u0026nbsp;4c).\u003c/b\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec27\" class=\"Section2\"\u003e \u003ch2\u003e3.4 Data integration\u003c/h2\u003e \u003cp\u003eA high-level approach was used for data fusion by allowing the integration of results coming from the different blocks. Samples common to NMR and LC-MS platforms were included in the analysis (n\u0026thinsp;=\u0026thinsp;56, 13 progressor, 43 stable). All the possible combinations of block integration were tested. Finally, the most performant models were obtained by integration of NMR, Positive and Negative MS modes. A total of 31 variables, coming from the 3 different blocks, were integrated, and PCA and OSC-PLS models were performed \u003cb\u003e(see Supplementary Fig.\u0026nbsp;2 and Supplementary Table\u0026nbsp;1)\u003c/b\u003e. The discrimination between the two groups, that was already visible through the PCA score plot was highlighted by the high discriminant performance of OSC-PLS model (Q2\u0026thinsp;=\u0026thinsp;0.829, p-value\u0026thinsp;=\u0026thinsp;0.00214) \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003ea\u003cb\u003e)\u003c/b\u003e. Like previously done for each separate dataset, a multivariate ROC curve was performed \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003eb\u003cb\u003e).\u003c/b\u003e The degree of separability was evaluated (20 variables, AUC\u0026thinsp;=\u0026thinsp;0.816, 95% Cl 0.651\u0026ndash;0.947, accuracy\u0026thinsp;=\u0026thinsp;70.2%) by showing a more performant power compared to models of single blocks. Only 10 \u0026ldquo;stable\u0026rdquo; and 2 \u0026ldquo;progressor\u0026rdquo; patients were misclassified in confusion matrix by showing a general increased classification power when coupling NMR and LC-MS selected variables (accuracy\u0026thinsp;=\u0026thinsp;0.81, precision\u0026thinsp;=\u0026thinsp;0.52, recall\u0026thinsp;=\u0026thinsp;0.85) \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003ec\u003cb\u003e).\u003c/b\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eBy comparing the performance of models from individual platforms and their combinations the integrated model incorporating NMR, both positive and negative modes (\u003cb\u003erefer to Table\u0026nbsp;5\u003c/b\u003e), exhibited the highest efficiency (AUC\u003csub\u003eNMR, POS, NEG\u003c/sub\u003e= 0.816 \u003cem\u003eversus\u003c/em\u003e AUC\u003csub\u003eNMR\u003c/sub\u003e= 0.794). Specifically, when examining the Q2 values of OSC-PLS models, only the model derived from the Positive mode outperformed the one integrating all 3 platforms. However, despite this, the p-value from the permutation test was lower in the integrated approach, suggesting that the observed result is unlikely to have occurred by random chance alone. Notably, considering the AUC value of the multivariate ROC curve, and the accuracy of the confusion matrix, the integration of NMR and the dual modes LC-MS platform yielded the most robust model.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabd\" border=\"1\"\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePlatform\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOSC-PLS (Q\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePermutation test (P-value)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eROC curve\u003c/p\u003e \u003cp\u003e(AUC)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eCI 95%\u003c/p\u003e \u003cp\u003e(AUC)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eConfusion matrix (accuracy)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNMR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.775\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.03617\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.794\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.650\u0026ndash;0.929\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.64\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePositive mode (LC-MS)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.933\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.03073\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.676\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.416\u0026ndash;0.875\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.71\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNegative mode (LC-MS)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.719\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.5381\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.752\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.596\u0026ndash;0.910\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.77\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eNMR, LC-MS Positive\u003c/b\u003e,\u003c/p\u003e \u003cp\u003e\u003cb\u003eLC-MS Negative\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003e0.829\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003e0.00214\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u003cb\u003e0.816\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e0.651\u0026ndash;0.947\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e\u003cb\u003e0.81\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNMR, LC-MS Positive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.768\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.04432\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.767\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.507\u0026ndash;0.907\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.66\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNMR, LC-MS Negative\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.783\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.0196\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.812\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.578\u0026ndash;0.954\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.70\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLC-MS Positive, LC-MS Negative\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.768\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.00618\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.767\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.5-0.919\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.64\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e\u003cp\u003e\u003cstrong\u003eTable\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e5\u003c/strong\u003e Comparison of models\u0026apos; performance for integrated and non-integrated models. In the table Q2 value for OSC-PLS models, permutation test p-values a, AUC of multivariate curves, CI 95% of multivariate \u0026nbsp;ROC curves, and accuracy of confusion matrix are reported.\u0026nbsp;\u003c/p\u003e \u003cp\u003e All the 31 annotated metabolites were used to generate a pathway analysis to explore the metabolic pathways impacted that could be correlated to the kidney function decrease. In particular, the result of this analysis highlights 11 metabolic pathways be involved in renal function decline with a p-value lower than 0.05 threshold (\u003cb\u003eTable\u0026nbsp;6).\u003c/b\u003e\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabe\" border=\"1\"\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePathways\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTotal\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHits\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ep-value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAminoacyl-tRNA biosynthesis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e48\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.69E-07\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePhenylalanine metabolism\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.0001256\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePhenylalanine, tyrosine and tryptophan biosynthesis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.00067098\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTryptophan metabolism\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e41\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.00078284\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGlyoxylate and dicarboxylate metabolism\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e32\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.0044684\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHistidine metabolism\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.012418\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePantothenate and CoA biosynthesis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.017357\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGlutathione metabolism\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.036209\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAlanine, aspartate and glutamate metabolism\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.036209\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePorphyrin and chlorophyll metabolism\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.041138\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGlycine, serine and threonine metabolism\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.048981\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable border=\"1\"\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e\u003cp\u003e\u003cstrong\u003eTable\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e6\u003c/strong\u003e significant metabolic pathways impacted by metabolites coming from the platforms\u0026rsquo; integration. (\u003cstrong\u003eTotal\u003c/strong\u003e= total number of metabolites composing the pathway, \u0026nbsp;\u003cstrong\u003eHits\u003c/strong\u003e= number of metabolites identified in our model)\u003c/p\u003e"},{"header":"4 Discussion","content":"\u003cp\u003eThis study aims to investigate the potential of metabolomics to predict the decline of measured renal function in KTRs at one year post KTx. We utilized a pilot cohort of well-characterized KTRs with detailed renal function assessments. Access to measured GFR, the gold standard for assessing renal function, clearly represents a key strength of this study. Urine samples were collected 3 months post-transplant, and glomerular filtration rate (GFR) was measured at both 3- and 12-months post-transplant. Based on these mGFR values, patients were categorized into two groups: those with stable renal function and those with 7% reduced mGFR. This 7% number is not arbitrary but based on well-established concepts, like \u0026ldquo;critical difference\u0026rdquo; or \u0026ldquo;least significant change (LSD)\u0026rdquo;, corresponding to clinically relevant changes in mGFR. To achieve a comprehensive metabolome coverage, we employed a multi-platform analytical strategy. 3 datasets (NMR, MS\u0026thinsp;+\u0026thinsp;and MS-) were thus obtained, analyzed and combined.\u003c/p\u003e \u003cp\u003eThe multivariate statistical analysis across various platforms allowed us to identify potential biomarkers associated with kidney function decline after transplantation. Notably, the classification of patients using confusion matrices demonstrated high performance and accuracy before metabolite identification. However, many discriminating metabolites used in these models were influenced by external factors, such as medication intake or environmental exposures (e.g., pesticides), rather than renal physiology. Given that patients in this cohort generally follow a similar standard treatment regimen with variations in dosages and comorbidities (\u003cb\u003esee Supplementary Table\u0026nbsp;2\u003c/b\u003e), we subsequently excluded features linked to exogenous compounds, unknown metabolites, and redundant biological substances from the predictive models. At this point, the selected features were used to define predictive and discriminant models for single and integrated datasets. By evidence, the integration of NMR and MS (in positive and negative mode) platforms allowed the model to gain in accuracy for groups prediction. The examination of both the OSC-PLS model and the confusion matrix scores clearly demonstrated the superior discrimination and classification performance of the combined model compared to the single-platform models. Moreover, through the identification of the features on which these models were based, it was possible to delineate a panel of putative biomarkers able to describe and predict kidney function decline from 3 to 12 months post KTx. Even if only Hippurate was found as significant in both platforms, several other common metabolites reached VIP scores. Furthermore, it was shown how metabolites coming from LC-MS and NMR techniques were sharing the same metabolic pathways allowing us to provide complementary information and a global view on metabolome of \u0026ldquo;progressor\u0026rdquo; patients.\u003c/p\u003e \u003cp\u003eThis analysis highlights the fact that all the identified features derive from various metabolic pathways going from amino acids metabolism (serine, glycine, phenylalanine, valine, hippuric acid etc..) to gut microbiota metabolism (TMAO, choline, ) whose importance in CKD frame has been suggested in the last years \u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e. The existing link between decline in kidney function and some metabolites highlighted in this study have been demonstrated.\u003c/p\u003e \u003cp\u003eFor example, TMAO is a well-known metabolite in the context of kidney dysfunction that has been proposed as potential biomarker of CKD in the last decade\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e. Increase in plasmatic TMAO has been reported as strongly linked to CKD \u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e,\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003eand inversely associated with estimated GFR \u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e,\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. In our study the downregulated TMAO level could be explained by an impaired renal extraction with a consequent depletion of this methylamine in urine, as supported by others\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e,\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e. Similarly, DMA is another methylamine that has been found to be downregulated in the present study. This metabolite has already been described in other works as being correlated to medullary damage and acute rejection post KTx \u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e,\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e .\u003c/p\u003e \u003cp\u003eAmong the metabolites with higher concentrations in the progressor group were choline, glycine, and serine. Elevated choline levels might be linked to decreased TMAO concentration, as choline is metabolized by bacteria to produce TMAO. This increase in choline is consistent with previous studies and may indicate tubulointerstitial dysfunction\u003csup\u003e\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003eor accumulation due to increased TMAO formation. Glycine levels might rise from choline generated through phosphatidylcholine metabolism, contributing to its increased circulating concentration and extraction, as demonstrated in earlier research\u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e. Higher urinary serine concentrations were also observed in patients with declining kidney function over 12 months. Serine, a precursor in glycine biosynthesis and metabolism, has been previously found at elevated levels in the plasma of CKD patients\u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eOther well-known metabolites in kidney disease are represented by tryptophan and kynurenine. Tryptophane/kynurenine pathways has gained interest in the last years because of its role in acute injury prediction\u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e. In addition to this, several other studies demonstrated decrease in tryptophan related to eGFR in serum of patients with kidney dysfunction but no changes in urine biofluids were reported. Other studies reported upregulation of kynurenine or downregulation of serotonin (other tryptophan derived metabolite) according to eGFR impairment in kidney transplanted recipients \u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e,\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e\u003c/sup\u003e. These observations suggest changes in tryptophan/kynurenine pathways associated with kidney GFR impairment. In our study, we hypothesize that the diminution of tryptophan and kynurenine levels in patients with kidney function decline can be linked to major GFR impairment in those patient compared to \u0026ldquo;stable\u0026rdquo; one as also showed by Colas et al.\u003csup\u003e11\u003c/sup\u003e who demonstrated an urinary increase in levels of these metabolites in tolerant kidney transplanted recipients compared to non-tolerant ones.\u003c/p\u003e \u003cp\u003eUremic toxin levels in urine decrease with the severity CKD. In this study, we found that 3-methylhistidine and hippuric acid, significant across both platforms, were retained and consequently downregulated in the urine of patients with declining renal function. Hippuric acid, a well-known uremic toxin linked to the gut microbiome, shows reduced urine concentration due to dysregulated active renal tubular secretion\u003csup\u003e\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e. 3-Methylhistidine, associated with muscle protein breakdown, has been implicated in CKD progression and complications, as indicated by increased plasma levels in stage 3\u0026ndash;4 CKD patients \u003csup\u003e\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e,\u003c/sup\u003e. The observed downregulation of these metabolites in \"progressor\" patients aligns with existing literature, despite limited information on their behavior in urine samples.\u003c/p\u003e \u003cp\u003eThe primary strength of this study reposed on the high quality of clinical data linked to patients. Notably, to our knowledge, this cohort is among the few documented in the literature where the GFR for each patient is not merely estimated but accurately measured by a reference method. The inclusion of mGFR values for characterizing patients constitutes a significant advantage for a deep and precise understanding of kidney function status. Moreover, the longitudinal nature of the sample cohort has facilitated the establishment of an innovative and clinically relevant patient classification. By employing this novel stratification method, we were able to investigate kidney function evolution in the months following kidney transplantation (KTx) and identify a panel of metabolites that could serve as early predictors of renal function decline in kidney transplant recipients (KTRs). Notably, many of these metabolites have been previously identified or associated with chronic kidney disease (CKD).\u003c/p\u003e \u003cp\u003eAnother focal point of this study was represented by the platform combination and the data integration that allowed an increase of prediction model performance. Particularly, NMR quantification represented an asset for data normalization constituting a crucial step in urine samples analysis. Indeed, the importance of normalization on creatinine continues to represent an open and fierce debate \u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e,\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e,\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e. In the context of this study, several preliminary analyses were conducted before validating the normalization to the dilution factor value. Furthermore, the positive outcomes achieved using the NMR platform convinced us to apply the same normalization procedure to MS data. This approach yielded effective results, thereby validating the robustness of the normalization method.\u003c/p\u003e \u003cp\u003eThe synergy between these platforms became apparent in our study conclusion, and their application in parallel helped to overcome the primary limitation of NMR represented by its lower sensitivity. Hence, through the combination of NMR's quantitative aspect with MS\u0026rsquo;s higher sensitivity, a panel of predictive biomarkers elucidating the deterioration of kidney function was established and this will provide a more encompassing vision of its biochemical processes. From a biological perspective, our analysis reveals minimal overlap between platforms, with only hippurate being statistically significant on both. Other metabolites such as creatinine, mannitol, citrate, phenylalanine, tryptophan, and tyrosine appear in the full list of VIPs (data not shown) but did not show significant discrimination in univariate analysis. To illustrate the complementarity of the platforms, a pathway analysis was performed, demonstrating that while different metabolites were identified by each technique, they were all part of the same metabolic pathways.\u003c/p\u003e \u003cp\u003eFrom a clinical point of view, in the stratification process, the \u0026ldquo;progressor\u0026rdquo; group is formed based on the mGFR progression, but there is no available information regarding the cause of its decline. Further investigation is warranted to gain a comprehensive understanding of the pathophysiologic events associated with the reduction in GFR within the \u0026ldquo;progressor\u0026rdquo; group.\u003c/p\u003e \u003cp\u003eAnother limitation of this study is the sample size, which poses a significant constraint. While the number of enrolled patients may be deemed reasonable, the uneven distribution of samples across categories complicates the identification of reliable biomarkers for clinical application.\u003c/p\u003e \u003cp\u003eMoreover, the single-site collection and the lack of validation cohort limit the robustness of the result, which is critical for routine clinical implementation. Consequently, additional studies and cohorts are mandatory to validate our preliminary findings.\u003c/p\u003e \u003cp\u003eFrom an analytical perspective, 11 out of 56 samples were initially identified as outliers due to the diverse drug regimens, including immunosuppressants. However, since the relevant features for the study were not affected by drug-related regions, all 56 samples were included in the univariate analysis. Despite these limitations, the results generated by this innovative exploratory and pilot study and the metabolites identified as potential biomarkers demonstrate the value of metabolomics and represent a very interesting basis on which to develop more advanced models.\u003c/p\u003e"},{"header":"5 Conclusions","content":"\u003cp\u003eEarly identification of post-transplant renal function trajectories, particularly the ability to predict a decline, is critical for optimizing the management and treatment of transplant recipients. In this context, we hypothesized that metabolomics could serve as a valuable and insightful tool. Consequently, we conducted an exploratory study aimed at predicting adverse renal function outcomes in transplant patients. Despite certain limitations inherent in our methodology, primarily related to the modest cohort size, our findings demonstrate that it is feasible to develop predictive models for renal function decline at 12 months post-transplant by analyzing urine samples collected as early as 3 months post-transplant. On the analytical front, our study highlights the complementary strengths of NMR and MS platforms in metabolomic profiling. Several metabolites identified as potential early biomarkers of renal function degradation have been associated with relevant metabolic pathways implicated in renal pathologies such as CKD. These preliminary findings not only open new avenues for research in KTx but also suggest a potential advancement in clinical diagnostic tools. From a clinical perspective, integrating metabolomics into routine practice could enhance early detection and management of renal function decline. By incorporating predictive models based on early metabolite changes, clinicians could identify at-risk patients earlier and tailor interventions more effectively. This proactive approach may improve patient outcomes by enabling more timely adjustments to treatment regimens and personalized care strategies. Future studies incorporating comprehensive clinical data and validating these models through multi-center or larger cohort studies will be essential to confirm their clinical utility. If successful, these models could become integral to routine post-transplant monitoring, ultimately contributing to better long-term outcomes and more efficient management of kidney transplant patients.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003e\u003cstrong\u003eAuthor information\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors and Affiliations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical Metabolomics Group, Center for Interdisciplinary Research on Medicines (CIRM), University of Liege, Liege, Belgium\u0026nbsp;\u003c/strong\u003eArianna Cirillo\u0026nbsp;\u0026amp;\u0026nbsp;Pascal de Tullio\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDivision of Nephrology-Dialysis-Transplantation, University of Li\u0026egrave;ge, CHU de Li\u0026egrave;ge, Li\u0026egrave;ge, Belgium\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGuillaume Resimont, Fran\u0026ccedil;ois Jouret \u0026amp; Pierre Delanaye\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eOniris, INRAE, LABERCA, Nantes, France\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eJustine Massias\u0026nbsp;\u0026amp; Yann Guitton\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMetaboHUB-MELISA, MetaboHUB-ANR-11-INBS-0010, Oniris, INRAE, LABERCA, Nantes, France\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eJustine Massias\u0026nbsp;\u0026amp; Yann Guitton\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInterdisciplinary Group for Applied Genoproteomics (GIGA),\u003c/strong\u003e\u003cstrong\u003eCardiovascular Sciences, University of Li\u0026egrave;ge, Li\u0026egrave;ge, Belgium\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFran\u0026ccedil;ois Jouret\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eParis Public Hospital System, Renal Physiology Unit, Bichat Hospital Paris, France\u0026nbsp;\u003c/strong\u003eEmmanuelle Vidal-Petiot \u0026nbsp;\u0026amp; Martin Flamant\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eParis Cit\u0026eacute; University\u0026nbsp;and Sorbonne Paris North University, INSERM U1148, LVTS, F-75018 Paris, France\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEmmanuelle Vidal-Petiot \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDepartment of Nephrology-Dialysis-Apheresis, University Hospital Car\u0026eacute;meau, N\u0026icirc;mes, France\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePierre Delanaye\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAC and PT are respectively a FRIA grantee and a research director of the Fonds de la Recherche Scientifique-FNRS. FJ received support from the University of Li\u0026egrave;ge (Fonds Sp\u0026eacute;ciaux \u0026agrave; la Recherche, Fonds L\u0026eacute;on Fredericq) and the FNRS (Research Credits 2016 and 2019). AC received support from the University of Liege (Fonds L\u0026eacute;on Fredericq).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eContributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePT, FJ, PD, \u0026nbsp;EVP and MF conceived the study. GR, EVP, and MF identified the patients and collected the samples. AC drafted the manuscript. YG and JM performed LC-MS samples and data analysis. AC performed the metabolomics analysis and interpreted the results. PT, FJ, PD, \u0026nbsp;EVP and MF supervised the study. All members of the study group contributed to data collection and provided input into the analysis, writing, and final approval of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCorresponding author\u0026nbsp;\u003c/strong\u003eCorrespondence to Pascal de Tullio\u0026nbsp;\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eEthics and declaration\u003c/strong\u003e\u003c/h3\u003e\n\u003ch3\u003e\u003cstrong\u003eConflicts of interest\u0026nbsp;\u003c/strong\u003e\u003c/h3\u003e\n\u003ch3\u003eAll authors state that there is no conflict of interest.\u003c/h3\u003e\n\u003ch3\u003e\u003cstrong\u003eEthical approval\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eFor the human cohort, ethical approval for the use of urinary samples and associated metadata in this study was obtained from CEERB Paris Nord (Ethic agreement code: IRB00006477).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe metabolomics and metadata reported in this paper are available via Metabolight (https://www.ebi.ac.uk/metabolights/) study identifier [REQ20260109215937]\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eGlassock, R. J., Warnock, D. G., \u0026amp; Delanaye, P. (2017). The global burden of chronic kidney disease: estimates, variability and pitfalls. \u003cem\u003eNature Reviews Nephrology\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e, 104\u0026ndash;114.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, D. Q., et al. (2019). Identification of serum metabolites associating with chronic kidney disease progression and anti-fibrotic effect of 5-methoxytryptophan. \u003cem\u003eNature Communications\u003c/em\u003e, \u003cem\u003e10\u003c/em\u003e, 1476.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePort, F. K., Wolfe, R. A., Mauger, E. A., Berling, D. P., \u0026amp; Jiang, K. (1993). Comparison of survival probabilities for dialysis patients vs cadaveric renal transplant recipients. \u003cem\u003eJournal Of The American Medical Association\u003c/em\u003e, \u003cem\u003e270\u003c/em\u003e, 1339\u0026ndash;1343.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHorvat, L. D., Shariff, S. Z., \u0026amp; Garg, A. X. (2009). \u0026amp; for the Donor Nephrectomy Outcomes Research (DONOR) Network. Global trends in the rates of living kidney donation. \u003cem\u003eKidney International\u003c/em\u003e, \u003cem\u003e75\u003c/em\u003e, 1088\u0026ndash;1098.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMirzakhani, M., Shahbazi, M., Oliaei, F., \u0026amp; Mohammadnia-Afrouzi, M. (2019). Immunological biomarkers of tolerance in human kidney transplantation: An updated literature review. \u003cem\u003eJournal Cellular Physiology\u003c/em\u003e, \u003cem\u003e234\u003c/em\u003e, 5762\u0026ndash;5774.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFiehn, O. C., \u0026amp; Genomics (2001). Metabolome Analysis, and Biochemical Modelling to Understand Metabolic Networks. \u003cem\u003eComparative and Functional Genomics\u003c/em\u003e, \u003cem\u003e2\u003c/em\u003e, 155\u0026ndash;168.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNicholson, J. K., Lindon, J. C., \u0026amp; Holmes, E. (1999). Metabonomics\u0026rsquo;: understanding the metabolic responses of living systems to pathophysiological stimuli via multivariate statistical analysis of biological NMR spectroscopic data. \u003cem\u003eXenobiotica\u003c/em\u003e, \u003cem\u003e29\u003c/em\u003e, 1181\u0026ndash;1189.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePan, Z., \u0026amp; Raftery, D. (2007). Comparing and combining NMR spectroscopy and mass spectrometry in metabolomics. \u003cem\u003eAnalytical And Bioanalytical Chemistry\u003c/em\u003e, \u003cem\u003e387\u003c/em\u003e, 525\u0026ndash;527.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRiccio, S., et al. (2022). New Insights from Metabolomics in Pediatric Renal Diseases. \u003cem\u003eChildren\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e, 118.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePereira, P. R., et al. (2022). Metabolomics as a tool for the early diagnosis and prognosis of diabetic kidney disease. \u003cem\u003eMedicinal Research Reviews\u003c/em\u003e, \u003cem\u003e42\u003c/em\u003e, 1518\u0026ndash;1544.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eColas, L., et al. (2022). Urinary metabolomic profiling from spontaneous tolerant kidney transplanted recipients shows enrichment in tryptophan-derived metabolites. \u003cem\u003eeBioMedicine\u003c/em\u003e, \u003cem\u003e77\u003c/em\u003e, 103844.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDahabiyeh, L. A., et al. (2023). Metabolomics profiling distinctively identified end-stage renal disease patients from chronic kidney disease patients. \u003cem\u003eScientific Reports\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e, 6161.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKobayashi, T., et al. (2014). A metabolomics-based approach for predicting stages of chronic kidney disease. \u003cem\u003eBiochemical and Biophysical Research Communications\u003c/em\u003e, \u003cem\u003e445\u003c/em\u003e, 412\u0026ndash;416.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJouret, F., et al. (2016). Nuclear Magnetic Resonance Metabolomic Profiling of Mouse Kidney, Urine and Serum Following Renal Ischemia/Reperfusion Injury. \u003cem\u003ePlos One\u003c/em\u003e, \u003cem\u003e11\u003c/em\u003e, e0163021.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTitan, S. M., et al. (2019). Metabolites related to eGFR: Evaluation of candidate molecules for GFR estimation using untargeted metabolomics. \u003cem\u003eClinica Chimica Acta\u003c/em\u003e, \u003cem\u003e489\u003c/em\u003e, 242\u0026ndash;248.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGlavan, M. R. (2025). Targeted Analysis of Serum and Urinary Metabolites for Early Chronic Kidney Disease. \u003cem\u003eIJMS\u003c/em\u003e 26, 2862.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMeeusen, J. W., St\u0026auml;mmler, F., Dasari, S., Schiffer, E., \u0026amp; Lieske, J. C. (2022). Serum myo-inositol and valine improve metabolomic-based estimated glomerular filtration rate among kidney transplant recipients. \u003cem\u003eFront Med\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e, 988989.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDelanaye, P., Cavalier, E., Depas, G., Chapelle, J. P., \u0026amp; Krzesinski, J. M. (2008). New Data on the Intraindividual Variation of Cystatin C. \u003cem\u003eNephron. Clinical Practice\u003c/em\u003e, \u003cem\u003e108\u003c/em\u003e, c246\u0026ndash;c248.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJiang, L., Huang, J., Wang, Y., \u0026amp; Tang, H. (2012). Eliminating the dication-induced intersample chemical-shift variations for NMR-based biofluid metabonomic analysis. \u003cem\u003eThe Analyst\u003c/em\u003e, \u003cem\u003e137\u003c/em\u003e, 4209.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePeng, T., Royer, A. L., Guitton, Y., \u0026amp; Le Bizec, B. (2017). Dervilly-Pinel, G. Serum-based metabolomics characterization of pigs treated with ractopamine. \u003cem\u003eMetabolomics\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e, 77.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKuhl, C., Tautenhahn, R., B\u0026ouml;ttcher, C., Larson, T. R., \u0026amp; Neumann, S. C. A. M. E. R. A. (2012). An Integrated Strategy for Compound Spectra Extraction and Annotation of Liquid Chromatography/Mass Spectrometry Data Sets. \u003cem\u003eAnalytical Chemistry\u003c/em\u003e, \u003cem\u003e84\u003c/em\u003e, 283\u0026ndash;289.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLawson, T. N., et al. (2017). msPurity: Automated Evaluation of Precursor Ion Purity for Mass Spectrometry-Based Fragmentation in Metabolomics. \u003cem\u003eAnalytical Chemistry\u003c/em\u003e, \u003cem\u003e89\u003c/em\u003e, 2432\u0026ndash;2439.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchymanski, E. L., et al. (2014). Identifying Small Molecules via High Resolution Mass Spectrometry: Communicating Confidence. \u003cem\u003eEnvironmental Science And Technology\u003c/em\u003e, \u003cem\u003e48\u003c/em\u003e, 2097\u0026ndash;2098.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDieterle, F., Ross, A., Schlotterbeck, G., \u0026amp; Senn, H. (2006). Probabilistic Quotient Normalization as Robust Method to Account for Dilution of Complex Biological Mixtures. Application in \u003csup\u003e1\u003c/sup\u003e H NMR Metabonomics. \u003cem\u003eAnalytical Chemistry\u003c/em\u003e, \u003cem\u003e78\u003c/em\u003e, 4281\u0026ndash;4290.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiller, R. C., et al. (2004). Comparison of Specific Gravity and Creatinine for Normalizing Urinary Reproductive Hormone Concentrations. \u003cem\u003eClinical Chemistry\u003c/em\u003e, \u003cem\u003e50\u003c/em\u003e, 924\u0026ndash;932.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eComparison of Uncertainties Related to Standardization of Urine Samples with Volume and Creatinine Concentration (2004). \u003cem\u003eThe Annals of Occupational Hygiene\u003c/em\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/annhyg/meh019\u003c/span\u003e\u003cspan address=\"10.1093/annhyg/meh019\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e doi:10.1093/annhyg/meh019.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGagnebin, Y., et al. (2017). Metabolomic analysis of urine samples by UHPLC-QTOF-MS: Impact of normalization strategies. \u003cem\u003eAnalytica Chimica Acta\u003c/em\u003e, \u003cem\u003e955\u003c/em\u003e, 27\u0026ndash;35.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGavaghan, C. L., Wilson, I. D., \u0026amp; Nicholson, J. K. (2002). Physiological variation in metabolic phenotyping and functional genomic studies: use of orthogonal signal correction and PLS-DA. \u003cem\u003eFEBS Letters\u003c/em\u003e, \u003cem\u003e530\u003c/em\u003e, 191\u0026ndash;196.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBoccard, J., \u0026amp; Rudaz, S. (2014). Harnessing the complexity of metabolomic data with chemometrics: Metabolomic data analysis with chemometrics. \u003cem\u003eJ Chemometrics\u003c/em\u003e, \u003cem\u003e28\u003c/em\u003e, 1\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSpiteri, M., et al. (2016). Data fusion between high resolution 1H-NMR and mass spectrometry: a synergetic approach to honey botanical origin characterization. \u003cem\u003eAnalytical And Bioanalytical Chemistry\u003c/em\u003e, \u003cem\u003e408\u003c/em\u003e, 4389\u0026ndash;4401.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHsu, C. N., \u0026amp; Tain, Y. L. (2022). Chronic Kidney Disease and Gut Microbiota: What Is Their Connection in Early Life? \u003cem\u003eIJMS\u003c/em\u003e 23, 3954.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZixin, Y., et al. (2022). TMAO as a potential biomarker and therapeutic target for chronic kidney disease: A review. \u003cem\u003eFrontiers In Pharmacology\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e, 929262.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBain, M. A., Faull, R., Fornasini, G., Milne, R. W., \u0026amp; Evans, A. M. (2006). Accumulation of trimethylamine and trimethylamine-N-oxide in end-stage renal disease patients undergoing haemodialysis. \u003cem\u003eNephrology Dialysis Transplantation\u003c/em\u003e, \u003cem\u003e21\u003c/em\u003e, 1300\u0026ndash;1304.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBell, J. D., et al. (1991). Nuclear magnetic resonance studies of blood plasma and urine from subjects with chronic renal failure: identification of trimethylamine-N-oxide. \u003cem\u003eBiochimica et Biophysica Acta (BBA) - Molecular Basis of Disease\u003c/em\u003e, \u003cem\u003e1096\u003c/em\u003e, 101\u0026ndash;107.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStubbs, J. R., et al. (2016). Serum Trimethylamine-N-Oxide is Elevated in CKD and Correlates with Coronary Atherosclerosis Burden. \u003cem\u003eJournal of the American Society of Nephrology\u003c/em\u003e, \u003cem\u003e27\u003c/em\u003e, 305\u0026ndash;313.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePelletier, C. C., et al. (2019). Elevation of Trimethylamine-N-Oxide in Chronic Kidney Disease: Contribution of Decreased Glomerular Filtration Rate. \u003cem\u003eToxins\u003c/em\u003e, \u003cem\u003e11\u003c/em\u003e, 635.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKrueger, E. S., Lloyd, T. S., \u0026amp; Tessem, J. S. (2021). The Accumulation and Molecular Effects of Trimethylamine N-Oxide on Metabolic Tissues: It\u0026rsquo;s Not All Bad. \u003cem\u003eNutrients\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e, 2873.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHsu, C. N. (2018). Gut Microbiota-Dependent Trimethylamine N-Oxide Pathway Associated with Cardiovascular Risk in Children with Early-Stage Chronic Kidney Disease. \u003cem\u003eIJMS\u003c/em\u003e 19, 3699.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGartland, K. P., Bonner, F. W., \u0026amp; Nicholson, J. K. (1989). Investigations into the biochemical effects of region-specific nephrotoxins. \u003cem\u003eMolecular Pharmacology\u003c/em\u003e, \u003cem\u003e35\u003c/em\u003e, 242\u0026ndash;250.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLe Moyec, L., et al. (1993). Proton Nuclear Magnetic Resonance Spectroscopy of Urine and Plasma in Renal Transplantation Follow-Up. \u003cem\u003eNephron\u003c/em\u003e, \u003cem\u003e65\u003c/em\u003e, 433\u0026ndash;439.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRhee, E. P., et al. (2013). A Combined Epidemiologic and Metabolomic Approach Improves CKD Prediction. \u003cem\u003eJournal of the American Society of Nephrology\u003c/em\u003e, \u003cem\u003e24\u003c/em\u003e, 1330\u0026ndash;1338.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYamaguchi, Y., et al. (2021). Plasma metabolites associated with chronic kidney disease and renal function in adults from the Baltimore Longitudinal Study of Aging. \u003cem\u003eMetabolomics\u003c/em\u003e, \u003cem\u003e17\u003c/em\u003e, 9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eL\u0026ouml;b, S., \u0026amp; K\u0026ouml;nigsrainer, A. (2009). Role of IDO in Organ Transplantation: Promises and Difficulties. \u003cem\u003eInternational Reviews of Immunology\u003c/em\u003e, \u003cem\u003e28\u003c/em\u003e, 185\u0026ndash;206.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLandsberg, A., et al. (2018). Non-invasive staging of chronic kidney allograft damage using urine metabolomic profiling. \u003cem\u003ePediatric Transplantation\u003c/em\u003e, \u003cem\u003e22\u003c/em\u003e, e13226.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGoek, O. N., et al. (2013). Metabolites associate with kidney function decline and incident chronic kidney disease in the general population. \u003cem\u003eNephrology Dialysis Transplantation\u003c/em\u003e, \u003cem\u003e28\u003c/em\u003e, 2131\u0026ndash;2138.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeguchi, T., et al. (2005). Renal Clearance of Endogenous Hippurate Correlates with Expression Levels of Renal Organic Anion Transporters in Uremic Rats. \u003cem\u003eJournal Of Pharmacology And Experimental Therapeutics\u003c/em\u003e, \u003cem\u003e314\u003c/em\u003e, 932\u0026ndash;938.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMutsaers, H. A. M., et al. (2013). Optimized Metabolomic Approach to Identify Uremic Solutes in Plasma of Stage 3\u0026ndash;4 Chronic Kidney Disease Patients. \u003cem\u003ePlos One\u003c/em\u003e, \u003cem\u003e8\u003c/em\u003e, e71199.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang, Z. H., et al. (2016). Metabolomic Signatures of Chronic Kidney Disease of Diverse Etiologies in the Rats and Humans. \u003cem\u003eJournal Of Proteome Research\u003c/em\u003e, \u003cem\u003e15\u003c/em\u003e, 3802\u0026ndash;3812.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWaikar, S. S., Sabbisetti, V. S., \u0026amp; Bonventre, J. V. (2010). Normalization of urinary biomarkers to creatinine during changes in glomerular filtration rate. \u003cem\u003eKidney International\u003c/em\u003e, \u003cem\u003e78\u003c/em\u003e, 486\u0026ndash;494.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWarrack, B. M., et al. (2009). Normalization strategies for metabonomic analysis of urine samples. \u003cem\u003eJournal of Chromatography B\u003c/em\u003e, \u003cem\u003e877\u003c/em\u003e, 547\u0026ndash;552.\u003c/span\u003e\u003c/li\u003e \u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"metabolomics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"mebo","sideBox":"Learn more about [Metabolomics](http://link.springer.com/journal/11306)","snPcode":"11306","submissionUrl":"https://submission.nature.com/new-submission/11306/3","title":"Metabolomics","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"metabolomics, kidney transplant recipients, NMR, LC-MS, multiplatform approach, patient follow-up","lastPublishedDoi":"10.21203/rs.3.rs-8620894/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8620894/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eIntroduction\u003c/h2\u003e \u003cp\u003eKidney transplantation (KTx) provides the best therapeutic outcomes for patients with end-stage renal disease. However, long-term graft survival remains a major clinical challenge, and current biomarkers are insufficient to reliably predict post-transplant kidney function evolution. Identifying early predictors of renal function decline is therefore crucial to improve the monitoring and management of kidney transplant recipients (KTRs).\u003c/p\u003e\u003ch2\u003eObjectives\u003c/h2\u003e \u003cp\u003eThis pilot study aimed to investigate whether multiplatform urine metabolomics could identify early predictive biomarkers of renal function decline between 3 and 12 months post-KTx.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eA cohort of 56 French KTRs was recruited. Measured glomerular filtration rate (mGFR) was assessed at 3 (M3) and 12 (M12) months post-transplant, while urine samples were collected at M3. Patients were classified as \u0026ldquo;progressor\u0026rdquo; or \u0026ldquo;stable\u0026rdquo; based on a\u0026thinsp;\u0026ge;\u0026thinsp;7% decline or stability in mGFR over the 9-month follow-up period. Untargeted metabolomic profiling was performed on urine samples using complementary Nuclear Magnetic Resonance (NMR) and Mass Spectrometry (MS) platforms. Multivariate statistical analyses were then applied to identify metabolites associated with mGFR evolution.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eMultivariate modeling revealed putative urinary biomarkers associated with renal function trajectories. The strongest predictive performance was achieved using a combined model integrating both MS- and NMR-derived biomarkers, highlighting the complementarity of the two analytical approaches.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eDespite being conducted on a relatively small cohort, this exploratory study demonstrates that urinary metabolomics, particularly when combining NMR and MS datasets, holds promise as a predictive tool for renal function evolution in kidney transplant recipients. These findings support further validation in larger, independent cohorts.\u003c/p\u003e","manuscriptTitle":"Multiplatform urine metabolomics for non-invasive prediction of one-year renal function decline in kidney transplant recipients: a pilot study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-22 14:21:41","doi":"10.21203/rs.3.rs-8620894/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-04-09T13:18:18+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-09T02:43:29+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-03T07:35:28+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-31T10:06:50+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"64684490586948994215481319411532273468","date":"2026-03-23T12:20:28+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"160046200886991274147616522629802062962","date":"2026-03-23T08:51:37+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"250160510843593754803807586662745008874","date":"2026-03-21T11:34:52+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-02-05T08:05:27+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"175256682543861904590192461008350716699","date":"2026-01-26T13:55:38+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"295988008442891494436314332696624487842","date":"2026-01-21T13:35:08+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-01-20T15:28:06+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-01-17T03:02:02+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-01-17T03:01:10+00:00","index":"","fulltext":""},{"type":"submitted","content":"Metabolomics","date":"2026-01-16T16:10:55+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"metabolomics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"mebo","sideBox":"Learn more about [Metabolomics](http://link.springer.com/journal/11306)","snPcode":"11306","submissionUrl":"https://submission.nature.com/new-submission/11306/3","title":"Metabolomics","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"f5a60cf7-a5a6-4464-a1ee-32b4252010d4","owner":[],"postedDate":"January 22nd, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"in-revision","subjectAreas":[],"tags":[],"updatedAt":"2026-04-09T13:25:13+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-22 14:21:41","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8620894","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8620894","identity":"rs-8620894","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}