Urinary betaine/creatinine ratio enhances diagnostic accuracy of PLA2R-Ab in idiopathic membranous nephropathy

Abstract

Background

The sensitivity of serum anti-phospholipase A2 receptor antibody (PLA2R-Ab) in diagnosing idiopathic membranous nephropathy (IMN) remains suboptimal, underscoring the need for complementary biomarkers, particularly in seronegative cases. This study aimed to identify urine metabolic signatures associated with IMN to improve diagnostic insight.

Methods

Untargeted liquid chromatography–tandem mass spectrometry (LC-MS/MS) metabolomic profiling was performed on urine samples from IMN patients and healthy controls (HCs). Differential metabolites were identified using multivariate statistical analyses and further evaluated through metabolic network and pathway enrichment analyses. Key metabolites were subsequently validated using targeted metabolomics and ELISA in an extended cohort comprising 53 IMN patients (30 PLA2R-Ab positive and 23 PLA2R-Ab negative), 30 HCs, 23 patients with IgA nephropathy (IgAN), and 7 with minimal change nephropathy (MCN). Core metabolites were validated using 5-fold cross-validation.

Results

Untargeted metabolomics revealed 116 significantly altered metabolites in IMN patients. Enrichment analysis highlighted notable perturbations in pathways including Ferroptosis, Taurine and hypotaurine metabolism, and Glycine, serine, and threonine metabolism. Among eight quantitatively validated metabolites, urinary betaine was consistently elevated in IMN. The urinary betaine-to-creatinine ratio (Betaine/uCr) was significantly higher in the IMN group compared with HCs and IgAN patients (all p < 0.05), and was independent of common clinical parameters. Betaine/uCr alone discriminated IMN from HCs with an AUC of 0.83 and exhibited high specificity (96%) in distinguishing IMN from IgAN. Importantly, combining Betaine/uCr with PLA2R-Ab serology improved diagnostic performance (AUC = 0.91) compared to either marker alone. 5-fold cross-validation showed that the results were stable and reliable.

Conclusions

Urinary metabolomics uncovered characteristic metabolic disturbances in IMN, involving pathways related to podocyte injury and osmotic regulation. The betaine/uCr showed preliminary potential as an auxiliary diagnostic marker when combined with PLA2R-Ab, though its specificity required further investigation. Therefore, broader studies are needed to establish its clinical utility.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12882-025-04650-x.

Introduction

Idiopathic membranous nephropathy (IMN), also known as primary membranous nephropathy, is a major global cause of chronic kidney disease [1]. Advanced IMN is clinically and pathologically characterized by disturbances in protein and lipid metabolism, renal failure, and renal vein thrombosis [2–5]. Current therapeutic strategies are limited, and late-stage interventions often fail to reverse established systemic damage [6, 7]. Consequently, early and accurate diagnosis of IMN is critical for improving patient survival and quality of life [8].

The primary methods for early IMN diagnosis are renal biopsy and serum anti-phospholipase A2 receptor antibody (PLA2R-Ab) testing [9, 10]. Renal biopsy remains the diagnostic gold standard; however, its invasive nature frequently results in poor patient compliance during early disease stages [11]. Serum PLA2R-Ab has emerged as a valuable non-invasive diagnostic marker. Nevertheless, serum PLA2R-Ab is detectable in only approximately 70–80% of patients with IMN, meaning a significant proportion of patients with active disease are seronegative and at risk of missed diagnosis if this test is relied upon exclusively [10, 12]. Notably, other target antigens such as thrombospondin type-1 domain-containing 7 A (THSD7A), neural epidermal growth factor-like 1 (NELL − 1), and neural cell adhesion molecule 1 (NCAM1) have been identified and are detectable via commercial immunoassays [13–15]. Despite these advances, PLA2R-Ab remains the clinical cornerstone, thus identifying novel biomarkers is essential to overcome its diagnostic limitations.

Metabolomics enables high-throughput profiling of metabolites in biological samples and is widely used for biomarker discovery in metabolic diseases [16, 17]. For example, Chasapi et al. utilized NMR-based urinary metabolomics to identify five discriminatory metabolites distinguishing IgA nephropathy (IgAN) from IMN, confirming the diagnostic utility of this approach [18]. Further supporting this, recent studies reveal distinct metabolic signatures in IMN pathogenesis: Ye et al. reported downregulation of dehydroepiandrosterone sulfate (DHEAS) among 215 serum metabolites associated with immune dysregulation [19], while Qu et al. identified disrupted tryptophan-kynurenine pathways (elevated L-tryptophan and L-kynurenine) as key discriminators between IMN and IgAN (AUC >0.85) [20]. These findings collectively characterize metabolic dysregulation as a hallmark of IMN. However, PLA2R-Ab negative cases remain diagnostically challenging due to the lack of validated biomarkers. To address this gap, we employed urinary metabolomics to identify potential disease-associated biomarkers, aiming to provide new methods and insights for the diagnosis and differential diagnosis of renal function impairment in IMN patients-with a particular focus on enhancing diagnostic precision for PLA2R-Ab negative cases and improving early detection sensitivity.

Methods

Study participants

A total of 113 samples were collected between April 2022 and December 2023, comprising 30 samples from healthy controls (HCs), 53 samples from IMN patients (including 30 from serum PLA2R-Ab positive subgroup and 23 from serum PLA2R-Ab negative subgroup), 23 samples from IgAN cases, and 7 samples from minimal change nephropathy (MCN) cases. Hereafter, all mentions of “PLA2R-Ab positive or negative” in this study refer to serum PLA2R-Ab status. For untargeted metabolomic analysis, urine samples from 3 PLA2R-Ab positive IMN patients and 3 HCs were analyzed. Separately, a distinct cohort of 3 PLA2R-Ab positive IMN patients and 3 HCs (not overlapping with those in the untargeted analysis) underwent urine targeted metabolomics to preliminarily validate the differential metabolites identified in the discovery phase. Finally, all 113 samples were incorporated into the validation phase via enzyme-linked immunosorbent assay (ELISA).

Inclusion criteria specified that samples were from individuals aged 18–75 years, with diagnoses of IMN, IgAN, or MCN confirmed by renal biopsy and pathological examination. None of the included samples were from individuals who had used corticosteroids, immunosuppressants, or immune-modulating medications at the time of sample collection. HC samples were obtained from individuals undergoing routine health examinations during the same period, with no abnormalities in serological tests, urinalysis, CT scans, or ultrasound examinations; these HCs were age- and sex-matched to the combined group of IMN, IgAN, and MCN cases. Exclusion criteria for all participants included: presence of secondary nephropathies (such as those related to systemic infections, autoimmune diseases, malignancies, or drugs), diabetic kidney disease, other major systemic illnesses (autoimmune, endocrine, or hematologic diseases), pregnancy, lactation, or oral contraceptive use. Healthy control individuals were additionally excluded for any underlying conditions such as diabetes, hypertension, tumors, or chronic kidney disease. For patients with MN who were seronegative for PLA2R antibodies, a diagnosis of idiopathic MN (IMN) was made following rigorous exclusion of common secondary causes. Specifically, for the 23 seronegative patients, details of secondary cause exclusion, along with findings from renal biopsy immunofluorescence (IgG subclasses) and electron microscopy (deposit location), are provided in Supplementary_Table_1_Exclusion Criteria.

Serum PLA2R-Ab levels were measured using a commercial ELISA kit (EUROIMMUN, Germany). According to the manufacturer’s criteria, results were classified as negative (< 14 RU/mL), weak positive (14–20 RU/mL), or positive (≥ 20 RU/mL). All patients in the IMN-positive group were PLA2R-Ab positive, while HCs, PLA2R-Ab negative IMN patients, and those with other nephropathies were negative.

Sample collection and preparation

Fasting morning urine samples (~ 10 mL) were collected from eligible participants after they were diagnosed with the disease via renal biopsy but prior to the initiation of pharmacotherapy. Following centrifugation, 500 µL aliquots were transferred to labeled tubes and stored at -80 °C to prevent freeze-thaw cycles. Sample preparation for both metabolomic approaches followed similar protocols: Samples were thawed at 4 °C, vortexed, and mixed with pre-cooled methanol/acetonitrile/water solution (2:2:1, v/v). After vortexing and low-temperature ultrasonication, samples were centrifuged at 14,000 × g for 10 min at 4 °C. The supernatant was vacuum-dried, and the resulting residue was reconstituted in 70% methanol. After further vortexing and low-temperature ultrasonication, samples were centrifuged again at 14,000 × g for 10 min at 4 °C, with the supernatant collected for analysis.

Untargeted metabolomics

Untargeted metabolomic profiling employed high-throughput liquid chromatography-tandem mass spectrometry (LC-MS/MS) for comprehensive metabolite detection. Separation was achieved using an ExionLC AD ultra-high-performance liquid chromatography (UHPLC) system equipped with T3 and HILIC columns. Metabolite detection utilized an AB QTRAP mass spectrometer operating in triple quadrupole (Qtrap) scan mode. Metabolite identification was performed against the Metware Database (MWDB) using retention time (RT), ion pairs, and MS/MS spectral matching.

Targeted metabolomics

Targeted metabolomics enabled absolute quantification of specific metabolites identified during screening. Separation used an Agilent 1290 Infinity LC UHPLC system with HILIC and T3 columns. Detection employed an AB QTRAP mass spectrometer in multiple reaction monitoring (MRM) mode. Raw MRM data were processed using MultiQuant software to extract chromatographic peaks. Metabolite concentrations were calculated from standard curves after normalizing peak areas to internal standards.

ELISA validation

To establish the detection conditions of the ELISA kits and validate intra-assay precision, a pilot study was conducted, with detailed methods and results available in Supplementary_Document_2_ELISA Pilot Study. This pilot study confirmed the optimal detection conditions for the target analytes, and the coefficient of variation (CV%) for all indicators (Kynurenine, O-Phosphoethanolamine, Betaine) was below the kit-specified threshold (≤ 8%) (actual range: 1.9% − 5.9%), indicating high stability and low variability of the detection system.

Identified metabolites (Kynurenine, O-Phosphoethanolamine, Betaine) were validated using commercial ELISA kits (Nanjing Camilo Biological Engineering Co., Ltd.). Optical density (OD) was measured with a Multiskan FC microplate reader. Sample concentrations were interpolated from standard curves, adjusted for dilution factors, and normalized to urinary creatinine concentrations for final analysis.

Data analysis

Statistical analyses and metabolomic analyses were performed using SPSS 23.0, MetaboAnalyst 6.0, SIMCA version 14.1, OriginPro, the Metware Cloud platform, and R version 4.5.1, as detailed below.

Continuous data were analyzed in SPSS 23.0: normally distributed data are presented as mean ± standard deviation (x̄ ± s) and compared using Student’s t-test (for two groups) or one-way ANOVA (for multiple groups); non-normally distributed data are expressed as median (interquartile range) and compared using the Mann-Whitney U test (for two groups) or Kruskal-Wallis test (for multiple groups). Categorical variables were also analyzed in SPSS 23.0, using the Chi-square test or Fisher’s exact test as appropriate.

Multivariate analyses, including orthogonal partial least squares-discriminant analysis (OPLS-DA) and partial least squares-discriminant analysis (PLS-DA), were conducted using SIMCA version 14.1. Metabolomic pathway analysis and visualization (including volcano plots and clustering heatmaps) were performed using the Metware Cloud platform. Additionally, KEGG metabolic network mapping and betweenness centrality analysis were conducted in MetaboAnalyst 6.0.

The evaluation of diagnostic performance, including stratified 5-fold cross-validation and receiver operating characteristic (ROC) curve analysis, was conducted exclusively in R version 4.5.1. The dataset was divided into three predefined subgroups: healthy controls, PLA2R-Ab positive IMN, and PLA2R-Ab negative IMN. Through random stratified sampling, five balanced training–test folds were generated. For each fold, ROC curves were constructed for the betaine/uCr, PLA2R-Ab, and a logistic regression model combining both biomarkers. The area under the curve (AUC) was evaluated for statistical significance via bootstrapping.

Correlations were assessed in OriginPro: Pearson correlation coefficients were used for normally distributed data, while Spearman correlation coefficients were applied for non-normally distributed data. All statistical tests were two-sided, with significance defined as a p-value < 0.05.

Results

Participant characteristics

The study cohort comprised 113 participants, including 30 HCs, 53 IMN patients (30 PLA2R-Ab positive, 23 PLA2R-Ab negative), 23 IgAN cases, and 7 MCN cases. Detailed clinical characteristics of all participants are presented in Table 1, with complete data available in Supplementary_Table_3_Clinical Characteristics. There were no statistically significant differences in the distribution of gender (p = 0.40) and age (p = 0.06) among the groups, whereas all urinalysis parameters and serum biochemical markers exhibited statistically significant variations across the groups (p < 0.05).

Table 1.

Clinical data of subjects

Group	HC (n = 30)	IMN PLA2R + (n = 30)	IMN PLA2R - (n = 23)	IgAN (n = 23)	MCN (n = 7)	p-value
Male (n, %)	13 (43.3%)	20 (66.7%)	13 (56.5%)	14 (60.9%)	3 (42.9%)	0.40
Age (years)	47.07 ± 8.64	49.47 ± 10.87	52.00 ± 10.48	44.22 ± 8.02	45.57 ± 8.14	0.06
ALB (g/L)	45.07 ± 1.90	23.46 ± 5.46	28.47 ± 7.41	37.56 ± 6.00	18.06 ± 3.57	< 0.05
Urea (mmol/L)	4.99 ± 0.96	4.87 ± 0.96	5.63 ± 1.73	6.89 ± 3.97	5.31 ± 2.30	< 0.05
CREA (µmol/L)	66.67 ± 11.87	67.80 (58.08, 80.05)	63.92 ± 16.00	105.10 (73.80, 127.10)	65.80 ± 19.34	< 0.05
eGFR (mL/min/1.73m2)	102.86 ± 11.43	99.24 ± 13.38	99.31 ± 19.12	78.25 ± 29.44	108.94 ± 25.47	< 0.05
TG (mmol/L)	1.02 ± 0.38	3.45 ± 2.99	1.84 ± 0.86	2.26 ± 1.56	3.97 ± 1.48	< 0.05
CHOL (mmol/L)	4.55 ± 0.63	8.23 ± 2.32	6.27 ± 2.05	4.75 ± 1.07	9.73 ± 3.34	< 0.05
HDL (mmol/L)	1.53 ± 0.33	1.39 ± 0.43	1.39 ± 0.44	1.15 ± 0.42	1.44 ± 0.62	< 0.05
LDL (mmol/L)	2.74 ± 0.57	5.60 ± 1.84	4.21 ± 1.82	3.02 ± 0.72	8.03 ± 1.92	< 0.05
UACR (mg/g Cr)	20.21 ± 6.49	910.74 (510.85, 2867.35)	1510.75 (689.41, 4057.91)	866.19 (597.42, 1210.41)	2234.13 (1006.65, 5247.68)	< 0.05

Alb, Albumin; Crea, Creatinine; eGFR, Estimated glomerular filtration rate; TG, Triglycerides; Chol, Total Cholesterol; HDL, High-density lipoprotein; LDL, Low-density lipoprotein; UACR, urinary albumin-to-creatinine ratio. The estimated glomerular filtration rate (eGFR) was calculated using the 2009 Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation (unit: mL/min/1.73 m²), based on serum creatinine (Scr, in µmol/L), age, and gender. Males: 141 × min(Scr/80.4, 1)^(-0.411) × max(Scr/80.4, 1)^(-1.209) × 0.993^Age; Females: 141 × min(Scr/61.9, 1)^(-0.329) × max(Scr/61.9, 1)^(-1.209) × 0.993^Age × 0.742 (min = smaller value, max = larger value; 80.4 and 61.9 are sex-specific Scr conversion factors in µmol/L)

Urine untargeted metabolomics analysis

Untargeted metabolomic profiling via LC-MS/MS detected 1,068 urinary metabolites categorized by chemical class (Fig. 1). These metabolites were first normalized by urinary creatinine and used to construct an OPLS-DA model in SIMCA, with fitting parameters R²X = 0.897, R²Y = 1, and Q² = 0.997. A permutation test (n = 200 for all permutations) was performed to validate the model, yielding R² intercepts = (0, 0.99) and Q² intercepts = (0, 0.95). Here, R²X reflects the model’s ability to explain variance in metabolite data, R²Y quantifies the explanation of group classification, and Q² indicates predictive performance (generally acceptable if Q² >0.5); overfitting is indicated when the original model’s Q² is close to the Q² intercepts. The initial OPLS-DA model showed obvious overfitting, as its Q² (0.997) was close to the Q² intercepts. To address this overfitting, metabolites with a variable importance in projection (VIP) score < 1 were excluded, retaining 479 metabolites for a revised OPLS-DA model (Fig. 2A) with parameters R²X = 0.96, R²Y = 1, and Q² = 0.999. A permutation test was conducted for this revised model (Fig. 2B), revealing R² intercepts = (0, 0.999) and Q² intercepts = (0, 0.829); however, the original Q² remained close to the Q² intercepts, meaning overfitting persisted. Thereafter, a PLS-DA model was constructed (Fig. 2C), with fitting parameters R²X = 0.778, R²Y = 0.995, and Q² = 0.836. A permutation test was performed for this PLS-DA model (Fig. 2D), generating R² intercepts = (0, 0.776) and Q² intercepts = (0, 0.0393). This model was deemed robust and reliable: its Q² (> 0.5) was substantially higher than the Q² intercepts (ruling out overfitting), and its high R²Y confirmed strong group discriminative ability.

Fig. 1.

Metabolite classification analysis. Pie chart displaying the distribution of 1,068 detected urinary metabolites by chemical class, with detailed counts and percentages: Amino acids (343, 32.12%), Organic acids (201, 18.82%), Nucleotides (100, 9.36%), Fatty acids (90, 8.43%), Glycerophospholipids (80, 7.49%), Carbohydrates (70, 6.55%), Benzene (50, 4.68%), Heterocyclic compounds (49, 4.59%), and Others (85, 7.96%). The Others category includes: Bile acids (19), Coenzymes and vitamins (19), Hormones (14), Tryptamine, Cholines, and Pigments (10), SL (9), Alcohols and amines (8), Aldehydes, Ketones, and Esters (1), Pteridines and derivatives (1), and other metabolites (4)

Fig. 2.

Metabolomic model construction, validation, differential metabolite screening, and clustering analysis. (A) Revised OPLS-DA model constructed using 479 metabolites (after excluding those with VIP < 1), with fitting parameters R²X = 0.96, R²Y = 1, Q² = 0.999. (B) Permutation test for the revised OPLS-DA model, showing R² intercepts = (0, 0.999) and Q² intercepts = (0, 0.829). (C) PLS-DA model constructed to resolve overfitting, with fitting parameters R²X = 0.778, R²Y = 0.995, Q² = 0.836. (D) Permutation test for the PLS-DA model, generating R² intercepts = (0, 0.776) and Q² intercepts = (0, 0.0393). (E) Volcano plot of differential metabolites (screened via PLS-DA results, criteria: VIP ≥ 1, p < 0.05, FC ≥ 2 or ≤ 0.5); 116 significant metabolites were identified (110 upregulated, 6 downregulated in IMN patients). Each dot represents one metabolite: red dots = upregulated differential metabolites, green dots = downregulated differential metabolites, gray dots = detected but non-significant metabolites; betaine is labeled. (F) Clustering heatmap showing good intra-group sample reproducibility; red squares represent metabolites upregulated relative to the control group, green squares represent metabolites downregulated relative to the control group (darker red indicates higher upregulation degree, darker green indicates higher downregulation degree)

Based on the PLS-DA model results, differential analysis for the volcano plot of differential metabolites (Fig. 2E) [with criteria: VIP ≥ 1, p < 0.05, and fold change (FC) ≥ 2 or ≤ 0.5] identified 116 significant metabolites, among which 110 were upregulated and 6 were downregulated in IMN patients. The clustering heatmap further confirmed good intra-group sample reproducibility across all groups (Fig. 2F), validating the reliability of the differential metabolite screening results.

Subsequently, metabolic network analysis of these 116 differential metabolites was conducted using MetaboAnalyst, identifying 20 hub metabolites with the highest node connectivity in the network (Fig. 3A). These metabolites served as central mediators in interactions between various metabolic components and in maintaining the overall stability of the metabolic network. Based on this metabolic network analysis, the betweenness centrality of these 20 hub metabolites was further calculated (Fig. 3B). Betweenness centrality quantifies the capacity of a metabolite to function as a “functional bridge” connecting distinct modules in the network: metabolites with high betweenness centrality can mediate the flow of metabolites or regulatory signals between otherwise isolated metabolic modules, rendering them key nodes for coordinating dysregulated metabolic pathways in IMN.

Fig. 3.

Metabolic Network and Pathway Enrichment Analysis of Differential Metabolites in IMN. (A) Metabolite interaction network (MetaboAnalyst) illustrating the 20 hub metabolites with the highest node connectivity. Each dot represents one differential metabolite; dot color and size correspond to connectivity (redder/larger dots = higher connectivity, purple dots = lower connectivity). These hubs act as central mediators in metabolite interactions and network stability maintenance. (B) Betweenness centrality of the 20 hub metabolites. Metabolites with high betweenness act as bridges for signal/metabolite flow between network modules, potentially mediating cross-pathway communication in IMN. (C) KEGG pathway enrichment analysis (DA Score). All enriched pathways show positive DA Scores (overall upregulation), so the x-axis was adjusted to 0–1. DA Score reflects pathway upregulation magnitude; dots closer to the right side of the x-axis (higher DA Scores) indicate a greater overall upregulation magnitude of the corresponding pathways. Dot color corresponds to enrichment p-value (darker red = higher significance), dot size represents the number of enriched metabolites (screening criteria for differential metabolites: VIP ≥ 1, p < 0.05, FC ≥ 2 or FC ≤ 0.5)

KEGG pathway enrichment analysis was performed on these 116 differential metabolites to elucidate their involvement in IMN-related metabolic reprogramming, which revealed 20 significantly enriched pathways. As shown in Fig. 3C, the Differential Abundance (DA) score - reflecting the magnitude of pathway upregulation - varied significantly. Pathways such as Ferroptosis, Taurine and hypotaurine metabolism, and Glycine, serine and threonine metabolism exhibited relatively high DA scores, indicating a prominent upregulation trend. Dot color represented the enrichment p-value, with D-Amino acid metabolism, Cysteine and methionine metabolism, and Carbon metabolism showing strong significance (darker red dots). Dot size corresponded to the number of enriched differential metabolites (screened by VIP ≥ 1, p < 0.05, and FC ≥ 2 or ≤ 0.5); larger dots (e.g., in Biosynthesis of amino acids and Aminoacyl-tRNA biosynthesis) indicated more differential metabolites mapped to these pathways. In contrast, pathways like beta-Alanine metabolism and Neomycin, kanamycin and gentamicin biosynthesis showed lower DA scores, less significant p-values (lighter-colored dots), and/or fewer metabolites. Notably, Ferroptosis, Taurine and hypotaurine metabolism, and Glycine, serine and threonine metabolism - bolded in Fig. 3C - are among the most relevant to idiopathic membranous nephropathy (IMN) pathogenesis and betaine metabolism [20–22].

Targeted metabolomics identification

Targeted Metabolomics Identification of the 20 candidate metabolites revealed eight with statistically significant concentration changes following creatinine normalization (p < 0.05, Table 2). These included: L-Glutamic acid, O-Phosphoethanolamine, L-Cysteine, L-Kynurenine, S-Adenosylhomocysteine, Arachidonic acid, Glucose 6-phosphate, and Betaine.

Table 2.

Metabolites associated with IMN identified by integrated untargeted and targeted metabolomic profiling

HMDB	Compound	Targeted Metabolomics		Untargeted Metabolomics
		FC	p-value	FC	p-value
HMDB0000148	L-Glutamic acid	5.16	< 0.05	5.02	< 0.05
HMDB0000224	O-Phosphoethanolamine	7.21	< 0.05	6.92	< 0.05
HMDB0000574	L-Cysteine	3.92	< 0.05	6.81	< 0.05
HMDB0000684	L-Kynurenine	4.29	< 0.05	3.801	< 0.05
HMDB0000939	S-Adenosylhomocysteine	4.007	< 0.05	3.45	< 0.05
HMDB0001043	Arachidonic acid	13.59	< 0.05	37.77	< 0.05
HMDB0001401	Glucose 6-phosphate	17.58	< 0.05	10.33	< 0.05
HMDB0000043	Betaine	3.53	< 0.05	2.82	< 0.05
HMDB0000158	L-Tyrosine	2.91	0.06	2.99	< 0.05
HMDB0000161	L-Alanine	4.37	0.06	12.27	< 0.05
HMDB0000172	L-Isoleucine	4.23	0.05	3.97	< 0.05
HMDB0000191	L-Aspartic acid	2.98	0.06	5.45	< 0.05
HMDB0000216	Norepinephrine	1.13	0.83	4.40	< 0.05
HMDB0000696	L-Methionine	5.27	0.08	3.93	< 0.05
HMDB0000700	Hydroxypropionic acid	0.62	0.07	0.43	< 0.05
HMDB0000842	Quinaldic acid	1.00	0.99	2.94	< 0.05
HMDB0001645	L-Norleucine	3.59	1.00	3.831	< 0.05
HMDB0001906	2-Aminoisobutyric acid	6.32	0.09	4.34	< 0.05
HMDB0001983	5’-Deoxyadenosine	8.26	0.16	0.37	< 0.05
HMDB0000132	Guanine	0.99	0.99	0.03	< 0.05

The metabolites listed were identified through an integrated workflow. Initial discovery was performed via untargeted metabolomics (n = 3 per group), from which 20 hub metabolites were selected. Subsequent targeted metabolomic validation was conducted on independent samples (n = 3 per group). Statistical significance was determined by p-value < 0.05. Fold change (FC) is calculated as IMN/Control. All reported concentrations were creatinine-normalized

ELISA validation

A total of 113 participants were enrolled in this study, including 30 HCs, 53 IMN patients (30 PLA2R-Ab positive, 23 PLA2R-Ab negative), 23 IgAN cases, and 7 MCN cases. Based on the established pathophysiological mechanisms of kidney diseases [23–26], ELISA validation was performed for three biomarkers: betaine, kynurenine, and O-phosphoethanolamine (OPE). Urinary levels of these biomarkers in the five groups were normalized to urinary creatinine (uCr), and the comparative results are detailed in Table 3. Box plots further intuitively illustrate the pairwise comparison differences among groups (Fig. 4A-C).

Table 3.

Comparison of urinary biomarker levels across five groups

Group	Betaine/uCr (×10− 3 mg/gCr)	Kynurenine/uCr (×10− 3 mg/gCr)	OPE/uCr (×10− 3 mg/gCr)
HC	80.26 (57.47, 101.41)	0.28 (0.21, 0.43)	0.36 (0.28, 0.49)
IMN PLA2R-Ab +	179.12 (109.50,277.29)	0.33 (0.22, 0.56)	0.30 (0.23, 0.42)
IMN PLA2R-Ab -	116.05 (94.63, 292.20)	0.31 (0.18, 0.54)	0.33 (0.25, 0.49)
IgAN	83.22 (69.64, 128.44)	0.40 (0.26, 0.60)	0.39 (0.24, 0.57)
MCN	118.87 ± 43.40	0.38 ± 0.27	0.37 ± 0.25
p-value	< 0.05	0.93	0.38

Data are presented as median (interquartile range) for non-normally distributed variables and mean ± standard deviation (SD) for normally distributed variables

Fig. 4.

Multiple Analyses of ELISA Results. (A) Box plot of Betaine/uCr levels among different groups; (B) Box plot of Kynurenine/uCr levels among different groups; (C) Box plot of OPE/uCr levels among different groups; (* indicates a statistically significant difference between two groups, p < 0.05).(D) ROC curve analysis of Betaine/uCr for the IMN group vs. HC group (AUC = 0.83); (E) ROC curve analysis of Betaine/uCr for the IMN group vs. IgAN group (AUC = 0.78); (F) ROC curve analysis of Betaine/uCr for the IMN group vs. MCN group (AUC = 0.68); (G) ROC analysis of three detection strategies in the non-IMN group (comprised IgAN and MCN) and IMN group: Betaine/uCr alone (AUC = 0.76), UACR alone (AUC = 0.55), and their combination (AUC = 0.77); (H) ROC analysis of three detection strategies in the IgAN group and IMN group: Betaine/uCr alone (AUC = 0.78), UACR alone (AUC = 0.61), and their combination (AUC = 0.80); (I) ROC analysis of three detection strategies in the MCN group and IMN group: Betaine/uCr alone (AUC = 0.68), UACR alone (AUC = 0.65), and their combination (AUC = 0.68)

The Betaine/uCr ratio was significantly elevated in patients with IMN (both PLA2R-Ab positive and negative) compared to HC and the IgAN group (all p < 0.05; Fig. 4A). No significant difference was observed between the PLA2R-Ab positive IMN and MCN groups (p = 0.15), whereas the Betaine/uCr ratio was significantly higher in the PLA2R-Ab negative IMN group than in the MCN group (p < 0.05). There was no significant difference in the Betaine/uCr ratio between the PLA2R-Ab positive and negative IMN subgroups (p = 0.29). Conversely, neither Kynurenine/uCr nor OPE/uCr showed significant differences in any group comparisons (all p > 0.05; Fig. 4B-C).

These results suggest that Betaine/uCr serves not only as a differential biomarker for distinguishing IMN from HCs, but also exhibits high specificity (96%; Table 4) in discriminating IMN from IgAN. The lack of a statistically significant difference between the PLA2R-Ab positive IMN group and the MCN group may be attributed to the limited sample size of the MCN group (n = 7) - a consequence of the difficulty in collecting MCN patient samples that met the inclusion and exclusion criteria of this study. To validate this preliminary finding and enhance the reliability of the comparison between these two groups, subsequent studies with an expanded MCN sample size (and ideally a larger overall cohort) are warranted.

Table 4.

ROC curve analysis of Betaine/uCr ratio for discriminating between groups

Comparison	AUC (95% CI)	Cut-off (× 10− 3 mg/g Cr)	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)	p-value
IMN vs. HC	0.83 (0.74–0.92)	96.99	81	76	86	69	< 0.05
IMN vs. IgAN	0.78 (0.68–0.89)	165.3	57	96	97	49	< 0.05
IMN vs. MCN	0.68 (0.52–0.84)	171.5	55	100	100	23	0.12

AUC: Area under the ROC curve; 95% CI: 95% confidence interval; Cut-off: Optimal cut-off value for Betaine/uCr (unit: ×10^− 3 mg/g Cr); PPV: Positive Predictive Value; NPV: Negative Predictive Value; p < 0.05 indicates a statistically significant difference in diagnostic performance

ROC curve analysis was performed to assess the diagnostic efficacy of the urinary Betaine/uCr ratio (unit: ×10^− 3 mg/g Cr) for IMN, with an AUC > 0.8 defined as the criterion for “certain diagnostic value”. For this analysis, IMN patients (including PLA2R-Ab positive and PLA2R-Ab negative subgroups) were merged into a single IMN group, and the results are presented in detail in Table 4, with corresponding visualizations provided in Fig. 4D-F.

When discriminating IMN from HC, Betaine/uCr exhibited strong diagnostic performance: it achieved an AUC of 0.83 (95% confidence interval [CI]: 0.74–0.92) with a cut-off value of 96.99 × 10^− 3 mg/g Cr, accompanied by 81% sensitivity, 76% specificity, 86% positive predictive value (PPV), and 69% negative predictive value (NPV) (p < 0.05). For distinguishing IMN from IgAN, Betaine/uCr showed moderate diagnostic utility: the AUC was 0.78 (95% CI: 0.68–0.89) at a cut-off value of 165.3 × 10^− 3 mg/g Cr, with 57% sensitivity, 96% specificity, 97% PPV, and 49% NPV (p < 0.05), highlighting its high specificity for this discrimination. In contrast, when differentiating IMN from MCN, Betaine/uCr had limited diagnostic efficacy: the AUC was 0.68 (95% CI: 0.52–0.84) at a cut-off value of 171.5 × 10^− 3 mg/g Cr, with 55% sensitivity, 100% specificity, 100% PPV, and 23% NPV, and no statistical significance was observed (p = 0.12) - a result potentially attributed to the small sample size of the MCN group (n = 7) as noted earlier.

To evaluate the independent diagnostic value of urinary betaine/uCr in discriminating IMN from other proteinuric nephropathies, we performed ROC analyses comparing its performance against UACR. For these comparisons, the IMN group included both PLA2R-Ab positive and negative subgroups, while the non-IMN group comprised IgAN and MCN cases. The analyses, presented in Fig. 4G-I, demonstrated that in distinguishing IMN from the combined non-IMN group, betaine/uCr alone (AUC = 0.76) significantly outperformed UACR alone (AUC = 0.55), though their combination provided no substantial improvement (AUC = 0.77). Similarly, for discriminating IMN from IgAN specifically, betaine/uCr (AUC = 0.78) showed superior performance to UACR (AUC = 0.61), with the combined model achieving an AUC of 0.80. In contrast, both biomarkers exhibited limited efficacy in differentiating IMN from MCN, either alone (betaine/uCr AUC = 0.68; UACR AUC = 0.65) or in combination (AUC = 0.68). These findings indicate that urinary betaine/uCr possesses diagnostic value independent of UACR for distinguishing IMN from IgAN, but has limited utility in separating IMN from MCN.

Combined ROC analysis of Betaine/uCr and PLA2R-Ab in the HC and IMN groups (Fig. 5H) showed that PLA2R-Ab alone had a sensitivity of 57% and specificity of 100%, while Betaine/uCr alone reached 81% sensitivity and 76% specificity. The combination of both markers improved sensitivity to 69% while maintaining 100% specificity. AUC comparisons indicated that both Betaine/uCr alone (0.83) and the combination model (0.91) outperformed PLA2R-Ab alone (0.78), suggesting that a combined detection strategy may significantly enhance the diagnostic accuracy for IMN.

Fig. 5.

5-Fold Cross-Validation (Betaine/uCr, PLA2R-Ab) and Correlation Analysis. (A) The ROC of training data from 5-fold cross-validation using Betaine/uCr (Fold 1–5 AUC = 0.80–0.85); (B) The ROC of training data from 5-fold cross-validation using PLA2R-Ab (Fold 1–5 AUC = 0.77–0.79); (C) The ROC of training data from 5-fold cross-validation using Betaine/uCr + PLA2R-Ab (Fold 1–5 AUC = 0.89–0.93); (D) The ROC of testing data from 5-fold cross-validation using Betaine/uCr (Fold 1–5 AUC = 0.75–0.89); (E) The ROC of testing data from 5-fold cross-validation using PLA2R-Ab (Fold 1–5 AUC = 0.75–0.83); (F) The ROC of testing data from 5-fold cross-validation using Betaine/uCr + PLA2R-Ab (Fold 1–5 AUC = 0.83–0.96); (G) Spearman correlation matrix analyzing the associations between Betaine/uCr, PLA2R-Ab, Alb, Urea, Crea, eGFR, TG, Chol, HDL, LDL, UACR, and 24 h uTP. The color intensity of matrix blocks reflects the magnitude of correlation coefficients, with red representing positive correlation and blue representing negative correlation; * indicates a statistically significant correlation (p < 0.05); (H) ROC analysis of three detection strategies (PLA2R-Ab alone, Betaine/uCr alone, and their combination) in the HC group (n = 30) and IMN group (n = 53), aiming to optimize IMN diagnosis and clarify their respective and combined contributions to diagnostic efficacy. The results are as follows: PLA2R-Ab alone (sensitivity = 57%, specificity = 100%, AUC = 0.78); Betaine/uCr alone (sensitivity = 81%, specificity = 76%, AUC = 0.83); combined detection of PLA2R-Ab and Betaine/uCr (sensitivity = 69%, specificity = 100%, AUC = 0.91)

To clarify the diagnostic efficacy of Betaine/uCr, PLA2R-Ab, and their combined index, this study performed analysis using ROC curves. In the training set (Fig. 5A-C), the AUC of Betaine/uCr alone ranged from 0.80 to 0.85 (Fig. 5A), and that of PLA2R alone ranged from 0.77 to 0.79 (Fig. 5B). In contrast, the AUC of the combined index increased to 0.89–0.93 (Fig. 5C), suggesting that the combined index exhibited significantly better diagnostic discrimination in the training set than either single index. In the independent test set (Fig. 5D-F), the AUC of Betaine/uCr was 0.75–0.89 (Fig. 5D), and that of PLA2R was 0.75–0.83 (Fig. 5E). The combined index further improved the AUC to 0.83–0.96 (Fig. 5F), with AUCs exceeding 0.9 in some folds, indicating that the combined model still possessed excellent generalization ability and diagnostic accuracy on unseen datasets. In this part, the combined application of Betaine/uCr and PLA2R can significantly enhance the diagnostic efficacy for the target disease, demonstrating good stability and discrimination ability in both training and testing phases.

Correlation analysis revealed significant positive associations between PLA2R-Ab and lipid markers (Chol: r = 0.42; TG: r = 0.28; LDL: r = 0.32; all p < 0.05). In contrast, Betaine/uCr showed no significant correlations with any of the clinical parameters (e.g., Alb, Urea, Crea, eGFR, lipid profiles, UACR, uTP) (Fig. 5G), indicating that Betaine/uCr may act as an independent biomarker for IMN, not confounded by common clinical indices in this 113 - participant cohort.

Discussion

This study aimed to identify potential disease-associated biomarkers and provide new methods and insights for the diagnosis and differential diagnosis of renal function impairment in IMN patients. To this end, an integrated workflow combining untargeted metabolomics, targeted validation, and ELISA was employed to characterize the urinary metabolic profile of patients with IMN, explore the underlying metabolic disturbance mechanisms of IMN, and evaluate candidate diagnostic indicators. A total of 113 participants were enrolled, including HC, IMN patients (subdivided into PLA2R-Ab positive and PLA2R-Ab negative subgroups), IgAN patients, and MCN patients. Analysis of participant characteristics revealed no significant differences in gender (p = 0.40) or age (p = 0.06) across groups, eliminating potential confounding from these demographic factors; in contrast, urinalysis parameter (e.g., UACR) and serum biochemical markers (e.g., ALB, CHOL, CREA) exhibited significant inter-group variations (p < 0.05). These findings align with IMN’s well-established clinical phenotype-characterized by massive proteinuria, hypoalbuminemia, and hypercholesterolemia driven by podocyte injury-which is consistent with descriptions of IMN pathophysiology in the study [27, 28].

Untargeted metabolomic profiling via LC-MS/MS detected 1,068 urinary metabolites, with amino acids (343 metabolites, 32.12%) and organic acids (201 metabolites, 18.82%) accounting for the largest proportions. This distribution reflects the critical role of the kidney in amino acid reabsorption [29] - under normal conditions, the majority of filtered amino acids are reabsorbed by renal tubules [30] and suggests that IMN may disrupt this process. To explore such metabolic disruptions through reliable differential metabolite screening, robust model construction is essential for subsequent analyses. Initially, the OPLS-DA model exhibited overfitting (Q² = 0.997, close to Q² intercepts); referencing established metabolomic strategies for mitigating overfitting [31, 32] (e.g., excluding low-contribution metabolites to enhance model stability), we removed metabolites with VIP < 1 and reconstructed a PLS-DA model. This revised model yielded robust fitting parameters (R²X = 0.778, R²Y = 0.995, Q² = 0.836), whose reliability was further confirmed by permutation tests [R² intercepts = (0, 0.776), Q² intercepts = (0, 0.0393)], laying a solid foundation for accurate identification of differential metabolites.

Using the criteria of VIP ≥ 1, p < 0.05, and FC ≥ 2 or ≤ 0.5, a total of 116 significant differential metabolites were identified, with 110 upregulated and 6 downregulated in IMN patients. KEGG pathway enrichment analysis revealed that these differential metabolites were primarily concentrated in three key pathways: Ferroptosis, Taurine and hypotaurine metabolism, and Glycine-serine-threonine metabolism. These pathways are not only metabolically interconnected but also play distinct roles in the pathophysiological process of IMN.

Ferroptosis is an iron-dependent form of regulated cell death driven by lipid peroxidation [33]. Given the central role of podocyte injury in IMN, the enrichment of this pathway is of great significance [21]. The accumulation of Ferroptosis-related metabolites (e.g., arachidonic acid, which was upregulated 13.59 - fold in targeted validation) suggests that lipid peroxidation may contribute to podocyte loss in IMN: arachidonic acid, a polyunsaturated fatty acid, serves as a substrate for lipid peroxide generation, and its accumulation in IMN patients may amplify oxidative stress in podocytes, ultimately leading to ferroptotic cell death [34]. This is consistent with findings from Jo et al. [35], which demonstrated that PLA2R autoimmunity-associated immune complex-induced reactive oxygen species (ROS) production in podocytes drives podocyte phenotypic alterations and dysfunction-processes closely linked to Ferroptosis.

The enrichment of the Taurine and hypotaurine metabolism pathway highlights its critical role in renal osmotic homeostasis [36]. Taurine, an organic osmolyte, is actively reabsorbed by renal tubules to maintain the hypertonic environment of the renal medulla and protect renal cells from osmotic stress-induced damage [37, 38]. In IMN patients, the upregulation of taurine-related metabolites (consistent with the pathway’s high differential abundance score [39]) may represent a compensatory response to osmotic imbalance caused by massive proteinuria [40]: the loss of albumin in urine reduces plasma oncotic pressure [41], leading to interstitial edema and disrupted medullary osmolarity, which in turn impairs taurine reabsorption by tubular epithelial cells [42]. Previous studies have confirmed that impaired taurine metabolism is associated with the progression of chronic kidney disease (CKD) [42], as osmotic stress-induced tubular damage exacerbates the decline in renal function-this conclusion further supports the potential relevance of taurine metabolism dysregulation to the progression of IMN to CKD [19, 43].

The Glycine-serine-threonine metabolism pathway is essential for podocyte structure and function [44]. Glycine and serine are precursors for the synthesis of key components of the podocyte slit diaphragm, such as nephrin and podocalyxin [44–46]; dysregulation of this pathway (e.g., upregulation of glycine-related metabolites) may impair the synthesis or stability of these structural proteins, leading to slit diaphragm disruption and increased albumin permeability. This is consistent with previous research, which has shown that IMN-associated immune activation impairs the expression of podocyte phenotypic markers (e.g., ZO-1, WT1)-changes that may be further aggravated by impairments in amino acid metabolism, a process essential for sustaining podocyte differentiation [35, 47].

Among the 20 hub metabolites identified via metabolic network analysis, betaine emerged as a key candidate following targeted validation (FC = 3.53 in targeted metabolomics, p < 0.05) and ELISA confirmation. To understand the biological basis of elevated urinary betaine/uCr ratios in IMN patients, a twofold mechanism rooted in renal physiology and IMN pathophysiology is proposed, supported by evidence from the current study. First, glomerular filtration barrier dysfunction in IMN may increase betaine filtration: damage to the slit diaphragm (evidenced by high UACR) [27, 48, 49] allows more freely filtered betaine to enter the tubular lumen [50, 51], exceeding the reabsorptive capacity of renal tubules. Second, tubular reabsorption of betaine may be impaired: previous studies have confirmed that downregulation of the betaine transporter SLC6A12 in CKD reduces tubular betaine reabsorption, leading to increased urinary excretion [52, 53]. Although SLC6A12 expression was not measured in the current study, the elevated betaine/uCr ratio in IMN patients may reflect similar transporter dysfunction-potentially induced by indirect effects of IMN, such as complement activation (e.g., C5b-9 deposition) [54] or pro-inflammatory cytokines (e.g., tumor necrosis factor-α) released by podocytes or infiltrating immune cells, which can damage tubular epithelial cells and reduce SLC6A12 expression. The lack of correlation between betaine/uCr and clinical parameters (e.g., eGFR, UACR) further suggests that betaine excretion is not merely a secondary effect of reduced glomerular filtration but may involve specific tubular dysfunction.

However, the proposition that renal tubules actively participate in IMN pathogenesis requires cautious interpretation. IMN is primarily characterized by immune complex deposition on podocytes, with podocytes being the principal target cells [55]. The elevated betaine/uCr ratio likely reflects glomerular-tubular crosstalk-where podocyte injury secondarily affects tubular function-rather than direct tubular involvement in the core pathology [56]. For instance, filtered proteins in IMN may induce tubular stress and transient transporter alterations (e.g., SLC6A12) [57]. In the absence of evidence for direct immune deposition or specific tubular injury markers, a primary role for tubules in IMN remains speculative.

To adequately address the specificity of urinary betaine in IMN, we must look beyond the shared pathway of glomerular–tubular crosstalk and focus on disease-specific pathophysiological amplifiers. The critical distinction lies in the nature of the podocyte injury: IMN is characterized by an antibody-driven, complement-mediated attack on podocytes [58, 59]. Its unique pathological signature comprises C5b-9-mediated podocyte injury, sustained complement activation [60–62], and the subsequent induction of tubular mitochondrial dysfunction and epigenetic dysregulation [63]. These processes collectively alter the expression and function of solute carriers [64], thereby disrupting betaine metabolism. This distinct pathophysiology is largely absent in MCN, which exhibits only transient tubular stress and minimal complement activation. As reported by Rahim Iranzad et al. [65], MCN lacks the significant podocyte structural disruption and sustained complement activation seen in FSGS and MN. This view is further corroborated by Masahiro Okabe et al. [66], in whom the %EGR1pod metric indicated the lowest degree of cellular stress in MCN patients. In contrast, FSGS is predominantly defined by a fibrotic phenotype. As emphasized in the study by [67], the pathological core of FSGS involves progressive glomerulosclerosis and interstitial fibrosis, which does not mechanistically overlap with the immune complex–driven pathology characteristic of MN. Therefore, the biomarker specificity of betaine arises from the systemic metabolic disturbance initiated by podocyte injury and uniquely perpetuated in IMN, as reflected in urinary excretion patterns.

Consistent with this pathophysiological basis, ELISA validation confirmed significantly elevated urinary betaine/uCr in IMN patients compared to HCs and IgAN patients (all p < 0.05). To further contextualize its diagnostic performance, we specifically evaluated betaine/uCr against UACR, the standard albuminuria marker. ROC analyses demonstrated that betaine/uCr consistently outperformed UACR in discriminating IMN from the combined non-IMN group and from IgAN specifically, with only marginal improvement observed upon their combination, indicating that its diagnostic value is independent of albuminuria levels. Betaine/uCr alone achieved an AUC of 0.83 for IMN vs. HC and 0.78 for IMN vs. IgAN. Notably, its combination with PLA2R-Ab improved diagnostic performance (AUC = 0.91), offering a key advantage for addressing the diagnostic gap in seronegative IMN. To address the limitation of the current absence of independent validation, 5-fold cross-validation was performed on three sets of models: Betaine/uCr, PLA2R-Ab, and their combination. The results demonstrated that all models exhibited stable performance, with the training sets and test sets yielding comparable results-indicating that no overfitting occurred.

Notwithstanding these findings, limitations exist. First, due to insufficient sample size, the ROC performance of betaine/uCr (including its combinatorial use with PLA2R-Ab) was not validated in an independent cohort-a critical step to prevent overfitting-induced inflation of diagnostic metrics and ensure generalizability. Second, serum betaine (which influences urinary excretion via glomerular filtration) was unmeasured; previous studies link serum betaine to urinary levels in chronic kidney disease (CKD), so unquantified serum levels may confound betaine/uCr interpretation. Third, patients with other proteinuric nephropathies (e.g., diabetic nephropathy, focal segmental glomerulosclerosis) were not evaluated, precluding assessment of betaine/uCr’s ability to distinguish IMN from other causes of proteinuria. Notably, prior work shows CKD/diabetic nephropathy patients also have reduced plasma betaine and increased urinary excretion-similar to IMN here-limiting betaine’s IMN specificity [25, 52, 68]. Additional limitations: MCN cohort (n = 7) may bias IMN-MCN discrimination, as evidenced by the relatively low diagnostic performance (AUC = 0.68), indicating limited discriminative value; cross-sectional design prevents linking betaine/uCr to IMN progression (e.g., longitudinal eGFR/proteinuria changes); no mechanistic experiments (e.g., renal SLC6A12 measurement, podocyte-tubular co-cultures) validated betaine elevation mechanisms.

Notably, the study has key strengths. It integrated untargeted/targeted LC-MS/MS metabolomics, ensuring robust differential metabolite identification. It included both PLA2R-Ab positive/negative IMN patients, addressing unmet needs across the IMN spectrum. Urinary metabolites offer non-invasive sampling-facilitating repeated collection and clinical translation. Finally, the use of five-fold cross-validation during model development enhances the reliability of our reported diagnostic performance metrics. Demonstrating that the combination of betaine/uCr and PLA2R-Ab improves diagnostic accuracy provides a practical combinatorial strategy. In conclusion, betaine/uCr shows promise as an auxiliary IMN marker (especially with PLA2R-Ab), but its specificity/generalizability needs confirmation in larger, diverse cohorts. Future studies should measure serum betaine and add mechanistic experiments. Despite limitations, the study’s rigorous metabolomic approach and focus on non-invasive biomarkers lay a valuable foundation for advancing IMN diagnostics.

Conclusion

This study revealed systematic urinary metabolic disturbances in IMN patients, implicating key pathways such as Ferroptosis and taurine, hypotaurine metabolism in podocyte injury and renal osmotic dysregulation. Targeted metabolomics and ELISA validation further revealed an elevated urinary betaine/uCr ratio, which, in combination with PLA2R-Ab, suggested preliminary auxiliary diagnostic value. However, the lack of independent cohort validation, unmeasured serum betaine, and non-specificity of betaine alterations in other proteinuric nephropathies currently preclude its use as an IMN-specific biomarker. Thus, these findings identify a disease-associated metabolic indicator rather than refute its clinical potential. Future studies should include broader cohorts, incorporate serum betaine measurement, and conduct longitudinal follow-up to clarify the diagnostic and prognostic relevance of betaine/uCr in IMN.

Supplementary Information

Below is the link to the electronic supplementary material.

Glossary

IMN

Idiopathic membranous nephropathy

PLA2R-Ab

Anti-phospholipase A2 receptor antibody

IgAN

IgA nephropathy

HC

Healthy Control

MCN

Minimal change nephropathy

LC-MS/MS

Liquid chromatography-tandem mass spectrometry

ROC

Receiver operating characteristic

AUC

Area under the curve

Alb

Albumin

Crea

Creatinine

eGFR

Estimated glomerular filtration rate

TG

Triglycerides

Chol

Total Cholesterol

HDL

High-density lipoprotein

LDL

Low-density lipoprotein

UACR

Urine Albumin-to-Creatinine Ratio

uTP

24-hour urinary total protein

CKD

chronic kidney disease

ELISA

Enzyme-linked immunosorbent assay

Author contributions

Conceptualization, investigation, visualization and writing: Jin Wang; writing - review & editing: Long Shao (Both Jin Wang and Long Shao contributed equally to this work as co-first authors); methodology and investigation: Baoxu Lin; software and validation: Ying Xi; data curation: Yangyang Jiang; conceptualization, supervision, project administration and funding acquisition: Xiaosong Qin. All authors read and approved the final manuscript.

Funding

This study was supported by Medical Masters Project of “Xingliao Talent Plan” in Liaoning Province, grant number No.YXMJ-LJ-09.

Data availability

The raw datasets analyzed in this study are available from the corresponding author upon reasonable request and with permission from the institutional review board. The raw untargeted metabolomics data have been deposited in the National Omics Data Encyclopedia (OMIX) database at the National Center for Bioinformation (China), under the accession code OMIX012074.

Declarations

Ethics approval and consent to participate

This retrospective study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Shengjing Hospital, China Medical University (Approval No. 2021PS781K). Fasting morning urine samples were collected from inpatients of the Department of Nephrology who strictly met the inclusion and exclusion criteria. All patient clinical data were de-identified to fully protect privacy. Notably, the samples used were residual specimens after routine clinical diagnosis, and the data were existing information from routine medical records. The study did not affect patients’ routine diagnosis and treatment: no additional tests/examinations were required for participation, and no harm was caused to patients. Thus, informed consent was waived by the Ethics Committee.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.