Potential diagnostic role of urinary exosomal microRNAs in immunoglobulin A nephropathy: A case-control study

Abstract

BACKGROUND

The most common type of primary glomerulonephritis is immunoglobulin A nephropathy (IgAN), which can eventually cause end-stage kidney disease among individuals. A kidney biopsy, which is invasive and has risks of mortality or morbidity, is currently the “gold standard” for diagnosis. Urinary exosomes contain abundant, well-preserved microRNAs (miRNAs), which are small, non-coding endogenous RNAs that may be used as non-invasive biomarkers. Studies on urinary exosomal miRNA profiles for the diagnosis of IgAN are rare.

AIM

To examine the profile of urinary exosomal miRNAs in Indian individuals diagnosed with IgAN.

METHODS

Over a period of 4 years (2020-2024), fifty biopsy-confirmed IgAN patients, fifty healthy controls, and fifty disease controls (DC) were recruited. Urinary exosomes were first discovered and then utilized for miRNA extraction. The nCounter^® Human v3 miRNA Expression Assay, a digital multiplex technique that evaluates 798 unique miRNA barcodes, was used to further analyze the extracted miRNAs. After the least absolute shrinkage and selection operator feature selection identified candidate miRNAs, logistic regression and the CombiROC algorithm were used.

RESULTS

The average age of patients diagnosed with IgAN was 36.32 years, with a standard deviation of 3.07 years. The average proteinuria was 2.69 ± 0.64 g/day, and the average creatinine level was 2.26 ± 0.318 mg/dL. Nine candidate miRNAs - hsa-miR-4532, hsa-miR-4488, hsa-miR-3158-3p, hsa-miR-151b, hsa-miR-3195, hsa-miR-1289, hsa-miR-20a-5p, hsa-miR-20b-5p, hsa-miR-32-5p, and hsa-miR-525-3p - successfully differentiate IgAN cases from both healthy and DC, exhibiting under the curve values of 0.7, 1, and 0.8 for healthy controls, DC, and IgAN cases, respectively. When compared to healthy and DC, a combination of just two miRNAs - hsa-miR-4532 and hsa-miR-548a-3p - was found to be sufficiently effective for detecting IgAN, with an area under the curve > 0.8.

CONCLUSION

Our investigation involving Indian participants revealed a marked alteration in the urinary exosomal miRNA patterns among individuals with IgAN compared to both healthy subjects and those with other kidney diseases, demonstrating the effectiveness of miRNAs in the non-invasive diagnosis of IgAN.

Key Words: Immunoglobulin A nephropathy; Urinary exosomal microRNA; Non-invasive biomarker; Indian cohort; Glomerulonephritis; Kidney biopsy

Core Tip: This case-control study evaluates urinary exosomal microRNAs (miRNAs) as noninvasive biomarkers for diagnosing immunoglobulin A nephropathy (IgAN), the most common primary glomerulonephritis. In biopsy-proven IgAN, healthy controls, and diverse glomerular disease controls, a NanoString-based urinary exosomal miRNA panel was profiled, followed by least absolute shrinkage and selection operator feature selection, logistic regression, and combinatorial receiver operating characteristic analysis. A focused set of IgAN-associated miRNAs, and a minimal multi-miRNA combination, demonstrated good diagnostic performance to distinguish IgAN from both healthy individuals and other glomerular diseases, suggesting a potential future role for urine exosomal miRNA signatures in reducing reliance on invasive kidney biopsy.

INTRODUCTION

Worldwide, immunoglobulin A nephropathy (IgAN) ranks as the most prevalent form of primary glomerulonephritis[1], with an incidence rate of approximately 2.1 per 100000 people, primarily affecting individuals in their twenties and thirties[2]. It progresses gradually, often manifesting as proteinuria, microhematuria, and a gradual decline in kidney function. Without early diagnosis and intervention, 15%-40% of patients may progress to kidney failure requiring dialysis within 10-20 years[3,4]. The definitive diagnosis of IgAN is typically made through a kidney biopsy, which is invasive and not easily repeatable due to the risk of complications like bleeding, hematuria, enlarged kidney, and so on. Kidney biopsy has complication rates up to 6.4%[5,6]. Small non-coding RNAs called microRNAs (miRNAs) are 20-25 nucleotides in length. They modulate gene activity via attaching to specific matching sequences on target mRNAs, inhibiting or degrading them. They are key in developing, detecting, and managing kidney diseases[7]. miRNAs are stable under various conditions - heat, changes in pH, long-term storage, and repeated freezing and thawing - because of their small size and protection within lipoprotein or lipid structures known as exosomes[8]. Combined with their non-invasive collection, these properties make miRNAs promising biomarkers for monitoring various diseases. However, there are challenges; miRNAs in urine are generally of low quality or degrade easily by endogenous RNAase, though those within urinary exosomes are protected from such degradation. Unlike kidney biopsies that sample only a small portion of tissue, urinary exosomes provide a comprehensive view of the urinary system[9,10].

Recent studies have explored using miRNA levels in urinary sediment[11] as non-invasive markers for IgAN. Research has shown changes in specific miRNAs in the urine of IgAN patients, correlating with kidney function or histological damage[12]. However, the data is limited and varies across different ethnic groups, with no specific data for the Indian population. Recognising that IgAN is prevalent and progressive among Asians, our recent study from the Indian population identified urinary exosomal miRNAs that could potentially differentiate “IgAN cases from healthy controls (HC)[13]. The purpose of the study is to compare the urinary exosomal miRNA profile of patients with IgAN” to that of disease and HC.

MATERIALS AND METHODS

Study design and population

From September 2020 to December 2024, the “Department of Nephrology at the Institute of Nephro-urology in Bengaluru” carried out a prospective clinical case-control observational study. The Institute of Nephro-urology’s Ethics Committee approved this study in keeping with the Declaration of Helsinki’s ethical principles (No. INU/IEC/CDSCO/11/20-21). Before being included in the study, each participant or guardian gave written informed consent. Urine samples were taken from 150 participants in the study to extract miRNA. Of 50 patients with biopsy-confirmed IgAN diagnosed throughout the study period made up the group. 50 volunteers between the ages of 18 and 65 who had normal kidney function, and the HC had no family or personal history of nephropathy or other comorbidities. To confirm normal kidney function in HC, urine tests and kidney function tests were performed (Table 1). We recruited 25 disease-control [disease controls (DC) or non-IgAN] patients with “biopsy-proven diabetic nephropathy (n = 9), lupus nephritis (n = 8), membranous nephropathy (n = 8), focal segmental glomerulosclerosis (n = 8), hypertensive nephrosclerosis (n = 9), or minimal change disease” (n = 8).

Table 1 Baseline characteristics of immunoglobulin A nephropathy cases and healthy controls.

Characteristics	IgA nephropathy (n = 50)	Healthy controls (n = 50)
Mean age (in years)	36.32 ± 3.06	39.81 ± 4.13
Male:female	3:1	3:1
Mean serum creatinine (mg/dL)	2.26 ± 0.31	0.8 ± 0.24
Mean eGFR (mL/minute/1.73 m2)	45.46 ± 8.49	133.5 ± 53.5
Mean proteinuria (g/day)	2.69 ± 0.64
M1	98%
E1	46%
S1	26%
T1	38%
T2	12%

IgA: Immunoglobulin A; eGFR: Estimated glomerular filtration rate.

Cases were prospectively identified, including consecutive patients whose kidney biopsies confirmed IgAN. A few days after the kidney biopsy result, a new urine sample was taken and kept at -80 °C. A prospective recruitment of HC was conducted. Healthy volunteers who were screened consisted of healthcare workers and patient caregivers. Subsequent kidney biopsies were used to prospectively identify the DC. Within a few days of the kidney biopsy result, a new urine sample was taken and kept at -80 °C. The study excluded people with associated problems such as diabetes, infections of the gastrointestinal, respiratory, or urinary systems, chronic liver diseases, systemic lupus erythematosus, or rheumatoid arthritis. Furthermore, the study excluded patients with IgA vasculitis or crescentic IgAN. To enable exosomal miRNA extraction, a freshly voided urine sample was collected from each participant after they or their legal guardians gave written authorization. Demographic or clinical data, including age, gender, 24-hour urinary protein excretion, or serum creatinine values, have been recorded for each IgAN patient in the trial at the time of their kidney biopsy. Glomerular filtration rate (eGFR) estimates were computed according to the 2021 chronic kidney disease epidemiology collaboration formula, utilizing creatinine values obtained during that year[14].

RNA extraction and Nanostring miRNA expression assay

The Norgen urine exosome RNA isolation Kit (No. Cat# 47200) (Norgen Biotek Corp., Thorold, Ontario, Canada) was used to extract exosomal RNA from human urine samples. To remove any sediment, urine samples were completely thawed and heated to 37 °C for 5 minutes. After that, RNA was concentrated using Zymo’s RNA Clean and Concentrator-5 (No. Cat# R1015) (Zymo Research Corp., Irvine, CA, United States) and eluted in 0.1 mL of elution buffer. Finally, 15 μL of the provided elution buffer was used to elute the RNA from the Zymo column. The Qubit RNA HS Assay Kit (Invitrogen, No. Cat# Q32855) (Thermo Fisher Scientific, Waltham, MA, United States) was used for quantification, and an Agilent 2100 Bioanalyzer equipped with a Pico chip was used to evaluate quality.

Using the ligation buffer and ligase included in the “NanoString nCounter Human v3 miRNA Expression Assay (NS_H_miR_v3b) kit (No. CSO-MIR3-12), miRNA (3 μL) was ligated to mir-Tag. The ligated product was denatured for 5 minutes at 85 ºC after being diluted with 15 μL of nuclease-free water. The 5 μL of this was hybridized overnight at 65 °C using Reporter and Capture” probes. MAN-C0009-07, the nCounter miRNA Expression Assay User Manual, was applied[15,16]. Following hybridization, samples were analyzed using a nanoString nCounter SPRINT device[17,18].

miRNA expression analysis

Each sample was evaluated using the nCounter Analysis System (from NanoString Technologies) and the nCounter Human v3 miRNA Expression Assay (NS_H_miR_v3b) panel, which has 798 distinct miRNA barcodes for endogenous miRNA[19]. The panel’s housekeeping genes are ACTB (beta-actin), B2M (beta-2-microglobulin), GAPDH (glyceraldehyde 3-phosphate dehydrogenase), RPL19 (ribosomal protein L19), or RPLP0 (ribosomal protein lateral stalk subunit P0). The panel comprises SpikeIn miRNAs, such as ath-miR159a from Arabidopsis thaliana, cel-miR-248 and miR-254 from Caenorhabditis elegans, and osa-miR414 and osa-miR442 from Oryza sativa, as well as positive and negative controls to evaluate test and ligation efficiency. The raw miRNA data that were generated as reporter code counts were further processed using the nSolver analysis software (NanoString Technologies, version 4.0). Imaging, binding density, positive control signals, detection limits, and ligation were all investigated as quality control measures before downstream analysis. Data has been normalized using the geometric mean of positive controls and the 100 most highly expressed miRNAs. The geometric mean of the top 100 most highly expressed genes was used to calculate these normalization variables.

The build ratio utility included in nSolver was employed to calculate the differential expression among groups (fold change). A significantly differentially expressed miRNA was defined as having a P-value of < 0.05 and a geometric mean expression of at least the average count of negative control probes in the panel for both the test and control groups. Upregulated or downregulated miRNAs have been defined via fold change ≥ 1.2 or < 1.2, respectively. Using the normalized expression data from the study’s samples, principal component analysis and heatmap creation were performed utilizing the ClustVis tool (Available from: http://biit.cs.ut.ee/clustvis/)[20]. The list of significantly differentially regulated miRNAs was used as input for over-representation analysis, which was carried out using the miEAA 2.0 (Available from: https://ccb-compute2.cs.uni-saarland.de/mieaa2/) online server for functional enrichment analysis[21]. Lists of miRNAs that were upregulated and downregulated were examined independently. The P-value adjustment method was set to false discovery rate (Benjamini-Hochberg) adjustment, the minimum number of hits per subcategory was set to two, and the significance level was set to 0.05 during the analysis. The over-representation study was conducted using the following databases: REACTOME (mirPathdb), KEGG, and the MNDR database for pathway and disease association enrichment, and mirPathdb for Gene Ontology Biological Process and Molecular Function enrichment.

miRNA feature selection utilizing least absolute shrinkage and selection operator regression

miRNAs that exhibited notable differences in expression among IgAN cases, HC, and DC, as detected by nSolver, have been analyzed using the “least absolute shrinkage and selection operator (LASSO)” regression to determine the most informative subset of features. Non-zero coefficients defined the LASSO-selected miRNA features, implemented using the glmnet package in R[22,23].

Diagnostic analysis of LASSO-selected miRNAs

Biomarker candidates were evaluated using a “logistic regression model with leave-one-out cross-validation, after randomly splitting the cohort into 70% training and 30% test sets; leave-one-out cross-validation was performed with the caret package in R. Receiver operating characteristic (ROC) curves and their corresponding area under the curve (AUC) values have been produced for” each candidate miRNA[8]. miRNAs with favourable AUCs in the logistic model were further tested in external datasets, using the present cohort as discovery and Gene Expression Omnibus datasets GSE141344 (renal biopsies) and GSE64306 (urinary sediments) as validation cohorts. Public datasets of urinary exosomal miRNAs in IgAN are not yet available. AUCs were computed with the pROC package, and CombiROC was applied to derive multi-marker panels for IgAN diagnosis[24-26].

Consensus clustering based on selected miRNAs

The capacity of LASSO-selected miRNAs to stratify samples was assessed by consensus clustering using the ConsensusClusterPlus package (v1.62.0)[27]. Euclidean distance and hierarchical clustering with Ward’s linkage (ward.D2) were used, with pItem = 0.80, pFeature = 1, and 500 resampling iterations. The most appropriate cluster count (k = 3) was identified using the cumulative distribution function, with the resulting clusters displayed using the heatmap package (v1.0.12)[28].

miRNA-mRNA-pathway network

For high-AUC miRNAs, experimentally validated mRNA targets were retrieved from miRTarBase v9.0[29,30]. Target gene functional and disease association enrichment was carried out via the DAVID bioinformatics resource (Available from: https://david.ncifcrf.gov/)[31]. A gradient bar plot of the top 20 enriched functions for up- and downregulated targets is provided in Supplementary Figure 1, and an integrated miRNA-mRNA-pathway–disease network has been constructed in Cytoscape (Available from: https://www.cytoscape.org/)[32].

Statistical analysis

All statistical analyses were performed utilizing the R software environment (version 4.0.2; Available from: https://www.r-project.org/). ROC analyses followed standard methods, and 2-sided P-values < 0.05 have been considered statistically significant.

RESULTS

In the IgAN cohort, the mean age was 36.32 ± 3.07 years, with 76% males and 24% females. Mean serum creatinine was 2.26 ± 0.32 mg/dL, proteinuria was 2.69 ± 0.64 g/day, and mean eGFR was 45.46 ± 8.49 mL/minute/1.73 m² (Table 1). The disease control cohort included individuals with confirmed diagnoses such as diabetic nephropathy, hypertensive nephrosclerosis, lupus nephritis (Class III/IV/V), minimal change disease, focal segmental glomerulosclerosis, and membranous nephropathy (Table 2).

Table 2 Baseline characteristics of disease controls.

Biopsy	Total number of cases	Mean age (in years)	Sex (male:female)	Mean serum creatinine (mg/dL)	eGFR (mL/minute/1.73 m2)	Proteinuria (grams/day)
Lupus nephritis	8	28.75 ± 5.1	1:3	1.34 ± 0.54	72.75 ± 12.23	1.67 ± 0.62
Diabetic nephropathy	9	47.2 ± 3.2	3:1	3.1 ± 0.43	18.66 ± 5.34	2.65 ± 1.21
Hypertensive nephropathy	9	52 ± 4.3	3:1	3.4 ± 1.1	16.8 ± 4.32	1.89 ± 87
Primary membranous nephropathy	8	39.5 ± 4.3	1:3	1.04 ± 0.56	83.5 ± 11.12	4.85 ± 2.32
Minimal change disease	8	10.25 ± 5.3	2:2	0.37 ± 0.48	165.25 ± 24.23	2.4 ± 1.22
Primary focal segmental glomerulosclerosis	8	18.75 ± 4.6	2:2	0.9 ± 0.63	126 ± 29.97	3.4 ± 1.12

eGFR: Estimated glomerular filtration rate.

About 137 miRNAs (35.2%) were significantly dysregulated only among IgAN cases and HC. About 215 miRNAs (55.3%) were significantly dysregulated only among IgAN cases, and other DC. About 37 miRNAs overlapped and significantly differentiated IgAN cases from HC as well as DC (9.5%) (Figure 1 and Supplementary Table 1). We would continue to call the 37 features IgAN-related features as they can potentially differentiate between IgAN and HC, as well as from other diseases. Principal component analysis of these overlapping 37 miRNA profiles (IgAN-related feature) significantly differentiated “IgAN cases from HC or DC” compared to the 798 genes panel in the nanostring technology (P-value = 0.0001 vs P-value = 0.0002) (Figure 2). The 37 overlapping miRNAs that could significantly differentiate IgAN cases from healthy or DC (Figure 1) were screened further using the LASSO regression model, which provided the 15 most important miRNAs (Table 3).

Figure 1 Venn plot: 137 microRNAs significantly differed between immunoglobulin A nephropathy cases and healthy controls. Of 215 microRNAs significantly differed between immunoglobulin A nephropathy cases and disease controls. It was found that 37 microRNAs commonly differentiated immunoglobulin A nephropathy cases from healthy controls and disease controls. IgAN: Immunoglobulin A nephropathy; HC: Healthy controls; DC: Disease controls.

Figure 2 Principal component analysis of 37 microRNA profiles shows that they can significantly differentiate between immunoglobulin A nephropathy cases, disease controls, and healthy controls compared to all the 798 microRNAs in the nanostring gene panel (left panel). PC: Principal component.

Table 3 Least absolute shrinkage and selection operator regression model showing 15 important microRNAs with non-zero coefficient values.

Healthy controls		IgAN		Disease controls
	Coefficient		Coefficient		Coefficient
Intercept	-2.277940703	Intercept	3.546568646	Intercept	-1.268627943
hsa-miR-409-5p	0.0763648402	hsa-miR-409-5p	-0.07957252572	hsa-miR-409-5p	0.003207685516
hsa-miR-4532	0.02020578069	hsa-miR-4532	-0.03750367821	hsa-miR-4532	0.01729789752
hsa-miR-320d	0.007678669715	hsa-miR-320d	-0.03569809053	hsa-miR-320d	0.02801942082
hsa-miR-3195	0.02502325322	hsa-miR-3195	-0.04476037204	hsa-miR-3195	0.01973711882
hsa-miR-664b-3p	0.005818441563	hsa-miR-664b-3p	-0.003261823244	hsa-miR-664b-3p	-0.002556618319
hsa-miR-151b	-0.008344659953	hsa-miR-151b	-0.02724506637	hsa-miR-151b	0.03558972632
hsa-miR-101-3p	0.0383626232	hsa-miR-101-3p	-0.008674373141	hsa-miR-101-3p	-0.02968825006
hsa-miR-492	-0.001133029441	hsa-miR-492	-0.0284297396	hsa-miR-492	0.02956276904
hsa-miR-511-5p	0.00200953104	hsa-miR-511-5p	1.32E-06	hsa-miR-511-5p	-0.002010848439
hsa-miR-769-5p	0.03050223486	hsa-miR-769-5p	0.01146667966	hsa-miR-769-5p	-0.04196891452
hsa-miR-499a-3p	0.02096262506	hsa-miR-499a-3p	0.01676192009	hsa-miR-499a-3p	-0.03772454516
hsa-miR-548n	-0.005970531682	hsa-miR-548n	0.01216365301	hsa-miR-548n	-0.006193121324
hsa-miR-654-3p	-0.05616509897	hsa-miR-654-3p	0.02554981477	hsa-miR-654-3p	0.0306152842
hsa-miR-29b-3p	-0.006355112674	hsa-miR-29b-3p	0.004412091123	hsa-miR-29b-3p	0.001943021551
hsa-miR-194-5p	-0.0000242413479	hsa-miR-194-5p	-1.72E-05	hsa-miR-194-5p	4.14E-05

IgAN: Immunoglobulin A nephropathy.

Out of 15 miRNAs, the logistic regression model could identify 4 markers with AUC values > 0.75 with a high ability to differentiate among HC or IgAN cases and 4 markers with AUC values > 0.75 to differentiate among DC or IgAN cases (Tables 4 and 5). Out of which, “hsa-miR-4532”, when validated within the public cohorts, was found to have the ability to differentiate IgAN from DC [AUC = 0.5764, 95% confidence interval (CI): 0.3399-0.8129; GSE141344] as well as HC (AUC = 0.8333, 95%CI: 0.6511-1; GSE64306), with a high AUC value. Following validation (Supplementary Table 2) within external cohorts, which identified multiple markers with potential AUC values, CombiROC analysis identified a combination of 3 miRNAs (hsa.miR.151b, hsa.miR.3195, hsa.miR.4532) that could significantly differentiate IgAN cases from DC with an AUC > 0.6 (Figure 3).

Figure 3 Workflow analysis. IgAN: Immunoglobulin A nephropathy; HC: Healthy controls; DC: Disease controls; miRNA: Micro ribonucleic acid; ORA: Over-representation analysis; MNDR: Mammal non-coding RNA-disease repository; LOOCV: Leave-one-out cross-validation; AUROC: Area under the receiver operating characteristic curve; LASSO: Least absolute shrinkage and selection operator; AUC: Area under the curve; ROC: Receiver operator curve.

Table 4 MicroRNAs with area under the curve values > 0.75 with a high ability to differentiate immunoglobulin A nephropathy cases from healthy controls.

miRNAs	hsa.miR.409.5p	hsa.miR.101.3p	hsa.miR.4532	hsa.miR.654.3p
Category	IgAN vs healthy
Accuracy (95%CI)	0.90 (0.58-0.99)	0.90 (0.58-0.99)	1 (0.71-1)	0.63 (0.30-0.89)
Sensitivity	0.50	0.50	1.00	1.00
Specificity	1.00	1.00	1.00	0.56
PPV	1.00	1.00	1.00	0.33
NPV	0.90	0.90	1.00	1.00
Prevalence	0.18	0.18	0.18	0.18
Detection rate	0.09	0.09	0.18	0.18
Detection prevalence	0.09	0.09	0.18	0.55
AUC (95%CI)	0.75 (0.26-1)	0.75 (0.26-1)	1 (1-1)	0.77 (0.60-0.94)
Kappa	0.62	0.62	1.00	0.31

miRNA: MicroRNA; CI: Confidence interval; PPV: Positive predictive value; NPV: Negative predictive value; AUC: Area under the receiver operating characteristic curve.

Table 5 microRNAs with area under the curve values > 0.75, with a high ability to differentiate immunoglobulin A nephropathy cases from disease controls.

miRNAs	hsa.miR.769.5p	hsa.miR.3195	hsa.miR.4532	hsa.miR.151b
Category	IgAN vs Disease control
Accuracy (95%CI)	0.8 (0.44-0.97)	0.8 (0.44-0.97)	0.8 (0.44-0.97)	0.8 (0.44-0.97)
Sensitivity	0.5	0.5	0.5	0.5
Specificity	1	1	1	1
PPV	1	1	1	1
NPV	0.75	0.75	0.75	0.75
Prevalence	0.4	0.4	0.4	0.4
Detection rate	0.2	0.2	0.2	0.2
Detection prevalence	0.2	0.2	0.2	0.2
AUC (95%CI)	0.75 (0.46-1)	0.75 (0.46-1)	0.75 (0.46-1)	0.75 (0.46-1)
Kappa	0.55	0.55	0.55	0.55

miRNA: MicroRNA; CI: Confidence interval; PPV: Positive predictive value; NPV: Negative predictive value; AUC: Area under the receiver operating characteristic curve.

Out of the 137 miRNAs (Figure 1) that significantly differentiated IgAN cases from HC, 47 miRNAs were found to be specific for IgAN vs HC, as they were not found to be dysregulated when DC were compared to HC (Figure 3). We would continue to call them features specific to IgAN, having the potential to differentiate from HC, but may not be the same as DC. Among these 47, LASSO regression selected 15 important miRNAs with biomarker potential (Figure 4), which were further used for checking their diagnostic value using LR (Table 6).

Figure 4 Least absolute shrinkage and selection operator coefficient plot showing 15 microRNAs that can differentiate immunoglobulin A nephropathy cases from healthy controls. miRNA: MicroRNA.

Table 6 Least absolute shrinkage and selection operator regression model selected 15 microRNAs that can differentiate immunoglobulin A nephropathy cases from healthy controls.

miRNA	hsa.miR.769.3p	hsa.miR.671.3p	hsa.miR.641	hsa.miR.612	hsa.miR.548ar.5p	hsa.miR.548a5p	hsa.miR.513c.3p	hsa.miR.4488	hsa.miR.3158.3p	hsa.miR.30e5p	hsa.miR.297	hsa.miR.222.3p	hsa.miR.219b3p	hsa.miR.1269b	hsa.miR.1268a
Accuracy (95%CI)	0.818181818181818 (0.482244147639827-0.97716880170004)	0.727272727272727 (0.390257440427579-0.939782265827093)	0.818181818181818 (0.482244147639827-0.97716880170004)	0.818181818181818 (0.482244147639827-0.97716880170004)	0.727272727272727 (0.390257440427579-0.939782265827093)	0.727272727272727 (0.390257440427579-0.939782265827093)	0.727272727272727 (0.390257440427579-0.939782265827093)	0.818181818181818 (0.482244147639827-0.97716880170004)	0.818181818181818 (0.482244147639827-0.97716880170004)	0.727272727272727 (0.390257440427579-0.939782265827093)	0.727272727272727 (0.390257440427579-0.939782265827093)	0.727272727272727 (0.390257440427579-0.939782265827093)	0.818181818181818 (0.482244147639827-0.97716880170004)	0.909090909090909 (0.587220083011617-0.997701027786186)	0.818181818181818 (0.482244147639827-0.97716880170004)
Sensitivity	0	0.5	0	0	0	0	0	0.5	0.5	0	0.5	0	0	0.5	0
Specificity	1	0.7777777778	1	1	0.8888888889	0.8888888889	0.8888888889	0.8888888889	0.8888888889	0.8888888889	0.7777777778	0.8888888889	1	1	1
PPV	NA	0.3333333333	NA	NA	0	0	0	0.5	0.5	0	0.3333333333	0	NA	1	NA
NPV	0.8181818182	0.875	0.8181818182	0.8181818182	0.8	0.8	0.8	0.8888888889	0.8888888889	0.8	0.875	0.8	0.8181818182	0.9	0.8181818182
Prevalence	0.1818181818	0.1818181818	0.1818181818	0.1818181818	0.1818181818	0.1818181818	0.1818181818	0.1818181818	0.1818181818	0.1818181818	0.1818181818	0.1818181818	0.1818181818	0.1818181818	0.1818181818
Detection rate	0	0.09090909091	0	0	0	0	0	0.09090909091	0.09090909091	0	0.09090909091	0	0	0.09090909091	0
Detection prevalence	0	0.2727272727	0	0	0.09090909091	0.09090909091	0.09090909091	0.1818181818	0.1818181818	0.09090909091	0.2727272727	0.09090909091	0	0.09090909091	0
AUC (95%CI)	0.5 (0.5-0.5)	0.638888888888889 (0.128164113234834-1)	0.5 (0.5-0.5)	0.5 (0.5-0.5)	0.444444444444444 (0.335557556414441-0.553331332474447)	0.444444444444444 (0.335557556414441-0.553331332474447)	0.444444444444444 (0.335557556414441-0.553331332474447)	0.694444444444444 (0.192500691940002-1)	0.694444444444444 (0.192500691940002-1)	0.444444444444444 (0.335557556414441-0.553331332474447)	0.638888888888889 (0.128164113234834-1)	0.444444444444444 (0.335557556414441-0.553331332474447)	0.5 (0.5-0.5)	0.75 (0.260009003864987-1)	0.5 (0.5-0.5)
Kappa	0	0.2325581395	0	0	-0.1379310345	-0.1379310345	-0.1379310345	0.3888888889	0.3888888889	-0.1379310345	0.2325581395	-0.1379310345	0	0.6206896552	0

miRNA: MicroRNA; CI: Confidence interval; PPV: Positive predictive value; NPV: Negative predictive value; AUC: Area under the receiver operating characteristic curve.

Of these 15 miRNAs, 3 miRNAs (hsa.miR.4488, hsa.miR.3158.3p, hsa.miR1269b) with an AUC of > 0.65 were selected (Table 6) for validation within external cohorts, where hsa.miR.4488 was observed with an AUC > 0.8. Following validation (Supplementary Table 2), the CombiROC algorithm identified a combination of 2 miRNAs (hsa.miR.4488 and hsa.miR.3158.3p) that could significantly differentiate among IgAN cases or HC with AUC: 0.815 (95%CI: 0.632-0.997). Of the 215 miRNAs that significantly differentiated IgAN cases from DC, we identified 11 important miRNAs using the LASSO algorithm, which were further tested for diagnostic accuracy using the LR model (Table 7). Out of these 11 miRNAs, 5 (hsa.miR.1289, hsa.miR.20a.5p.hsa.miR.20b.5p, hsa.miR.32.5p, hsa.miR.525.3p, hsa.miR.548a.3p) were identified with potential AUC values by the LR model (Table 7).

Table 7 Logistic regression statistics showing area under the curve of 11 important least absolute shrinkage and selection operator-selected microRNAs that can differentiate between immunoglobulin A nephropathy cases and disease controls.

miRNA	hsa.miR.1224.5p	hsa.miR.1279	hsa.miR.1289	hsa.miR.20a5p.hsa.miR.20b5p	hsa.miR.32.5p	hsa.miR.365b5p	hsa.miR.4787.3p	hsa.miR.520c.3p	hsa.miR.525.3p	hsa.miR.548a3p	hsa.miR.593.3p
Accuracy (95%CI)	0.8 (0.443904537692359-0.974789273673167)	0.8 (0.443904537692359-0.974789273673167)	0.9 (0.554983882971805-0.997471421455538)	0.9 (0.554983882971805-0.997471421455538)	0.9 (0.554983882971805-0.997471421455538)	0.8 (0.443904537692359-0.974789273673167)	0.8 (0.443904537692359-0.974789273673167)	0.8 (0.443904537692359-0.974789273673167)	0.9 (0.554983882971805-0.997471421455538)	1 (0.691502892181239-1)	0.8 (0.443904537692359-0.974789273673167)
Sensitivity	0.75	0.5	0.75	0.75	0.75	0.5	0.5	0.5	0.75	1	0.5
Specificity	0.833333333	1	1	1	1	1	1	1	1	1	1
PPV	0.75	1	1	1	1	1	1	1	1	1	1
NPV	0.833333333	0.75	0.857142857	0.857142857	0.857142857	0.75	0.75	0.75	0.857142857	1	0.75
Prevalence	0.4	0.4	0.4	0.4	0.4	0.4	0.4	0.4	0.4	0.4	0.4
Detection rate	0.3	0.2	0.3	0.3	0.3	0.2	0.2	0.2	0.3	0.4	0.2
Detection prevalence	0.4	0.2	0.3	0.3	0.3	0.2	0.2	0.2	0.3	0.4	0.2
AUC (95%CI)	0.791666666666667 (0.497218723153255-1)	0.75 (0.467103566480957-1)	0.875 (0.630004501932493-1)	0.875 (0.630004501932493-1)	0.875 (0.630004501932493-1)	0.75 (0.467103566480957-1)	0.75 (0.467103566480957-1)	0.75 (0.467103566480957-1)	0.875 (0.630004501932493-1)	1 (1-1)	0.75 (0.467103566480957-1)
Kappa	0.583333333	0.545454546	0.782608696	0.782608696	0.782608696	0.545454546	0.545454546	0.545454546	0.782608696	1	0.545454546

miRNA: MicroRNA; CI: Confidence interval; PPV: Positive predictive value; NPV: Negative predictive value; AUC: Area under the receiver operating characteristic curve.

Following validation within external cohorts, which provided multiple markers with AUC > 0.6 to differentiate between IgAN and DC (Supplementary Table 2), the CombiROC algorithm predicted a combination of 5 miRNAs (hsa-miR-1289,hsa-miR-20a-5p,hsa-miR-20b-5p,hsa-miR-32-5p,hsa-miR-525-3p) to differentiate IgAN from DC with better AUC than when using individual markers. Finally, considering the above combination of potential biomarker miRNAs, a new combination of 9 miRNAs was identified to significantly differentiate IgAN from DC as well as HC (Tables 8, 9 and 10).

Table 8 Regulation status of 9 microRNAs within multiple comparisons performed using the nCounter microRNA expression assay.

Type of differentiation	miRNAs	IgAN vs HC foldchange	IgAN vs HC P value	DC vs HC foldchange	DC vs HC P value	IgAN vs DC foldchange	IgAN vs DC P value
Differentiate IgAN and disease control	hsa-miR-1289	-1.19	0.05164487	-2.48	0.00001151	2.08	0.00001932
	hsa-miR-20a-5p+hsa-miR-20b-5p	-1.14	0.32179308	2.16	0.00004533	-2.46	0.00000026
	hsa-miR-32-5p	1.11	0.58471572	2.74	0.0002994	-2.48	0.0002138
	hsa-miR-525-3p	-1.13	0.46521613	2.05	0.00166705	-2.31	0.00012489
	hsa-miR-151b	-1.5	0.01793812	1.65	0.00405257	-2.48	0.00000064
	hsa-miR-4532	-2.06	0.00113097	1.17	0.68054599	-2.4	0.04225222
	hsa-miR-3195	-1.78	0.00082084	1.21	0.2488215	-2.14	0.00013383
Differentiate IgAN and healthy control	hsa-miR-4488	-2.65	0.02099009	-1.33	0.52975196	-2	0.05507494
	hsa-miR-3158-3p	-1.79	0.00653545	1.02	0.95068687	-1.82	0.07137911

miRNA: MicroRNA; IgAN: Immunoglobulin A nephropathy; HC: Healthy controls; DC: Disease controls.

Table 9 The combination of these 9 microRNAs has the potential to diagnose immunoglobulin A nephropathy cases with an area under the receiver operating characteristic curve > 0.8¹.

Discovery cohort with 9 miRNAs	hsa-miR-4532, hsa-miR-4488, hsa-miR-3158-3p, hsa-miR-151b, hsa-miR-3195, hsa-miR-1289, hsa-miR-20a-5p+hsa-miR-20b-5p, hsa-miR-32-5p, hsa-miR-525-3p
0 = Control, 1 = Disease, 2 = IgAN
Class	Control (0)	Disease (1)	IgAN (2)
Accuracy (95%CI)	0.7692 (0.4619-0.9496)	0.8462 (0.5455-0.9808)	0.7692 (0.4619-0.9496)
Sensitivity	0.33333	0.33333	0.8571
Specificity	0.9	1	0.6667
Positive predictive value	0.5	1	0.75
Negative predictive value	0.81818	0.83333	0.8
Prevalence	0.23077	0.2308	0.5385
Detection rate	0.07692	0.07692	0.4615
Detection prevalence	0.15385	0.07692	0.6154
AUC (95%CI)	0.733 (0.5-0.93)	1 (1-1)	0.88 (0.57-1)
Kappa	0.2642	0.4348	0.5301

¹Validation is presented in Supplementary Table 2, within the external cohort GSE64306.

miRNA: MicroRNA; IgAN: Immunoglobulin A nephropathy; CI: Confidence interval; AUC: Area under the receiver operating characteristic curve.

Table 10 A combination of 2 microRNAs could diagnose immunoglobulin A nephropathy cases with an area under the receiver operating characteristic curve > 0.8.

Discovery cohort with 2 microRNAs	hsa-miR-4532, hsa-miR-548a-3p
0 = Healthy controls, 1 = Disease controls, 2 = IgA nephropathy cases
Class	Control (0)	Disease (1)	IgAN (2)
Accuracy (95%CI)	0.7692 (0.4619-0.9496)	0.8462 (0.5455-0.980)	0.8462 (0.5455-0.9808)
Sensitivity	0.6667	1	1
Specificity	1	1	0.8333
PPV	1	1	0.875
NPV	0.9091	1	1
Prevalence	0.2308	0.2308	0.5385
Detection rate	0.1538	0.2308	0.5385
Detection prevalence	0.1538	0.2308	0.6154
AUC (95%CI)	0.8333333 (0.5-1)	1 (1-1)	0.9047619 (0.741-1)
Kappa	0	0.4348	0.6905

miRNA: MicroRNA; IgAN: Immunoglobulin A nephropathy; CI: Confidence interval; PPV: Positive predictive value; NPV: Negative predictive value; AUC: Area under the receiver operating characteristic curve.

Table 9 shows the combination of these 9 miRNAs that have the potential to diagnose IgA nephropathy cases with an AUC > 0.8. Validation is presented in Supplementary Table 2, within the external cohort GSE64306. Also, among these 9 miRNAs, a combination of just 2 miRNAs (hsa-miR-4532, hsa-miR-548a-3p) had an AUC of > 0.9 to potentially diagnose IgAN cases (Table 10). Table 11 shows that a combination of 2 miRNAs could diagnose IgAN cases with an AUC > 0.8. The combination of these 2 miRNAs was validated in the external cohort GSE64306 (Supplementary Table 3). Functional over-representation analysis was performed, separately for IgAN vs HC and DC vs HC. We observed the differences within the signalling pathways being enriched between IgAN and DC. Supporting gradient bar plot visualising the top 20 significant functions from upregulation and downregulation categories for both comparisons has been provided in Supplementary Figure 1.

Table 11 Status of the 5 published microRNA markers as observed from our previous study, within 3 comparisons (immunoglobulin A nephropathy vs healthy control, disease control vs healthy control, and immunoglobulin A nephropathy vs disease control).

IgA nephropathy vs healthy control
Probe name	Nephropathy	Control	Ig A nephropathy vs healthy control	P-value of: Nephropathy vs control
hsa-miR-146b-3p	4.07	13.3	-3.27	0.00047341
hsa-miR-221-5p	8.72	14.66	-1.68	0.00512936
hsa-miR-4532	7.01	12.3	-1.76	0.00655229
hsa-miR-599	7.26	16.63	-2.29	0.00649493
hsa-miR-664b-5p	3.85	9.47	-2.46	0.00083272
IgA nephropathy vs disease control
Probe name	Nephropathy	Disease control	IgA nephropathy vs disease control	P-value of: Nephropathy vs disease control
hsa-miR-146b-3p	6.17	8.4	-1.36	0.26787862
hsa-miR-221-5p	11.35	11.84	-1.04	0.73601639
hsa-miR-4532*	8.79	21.09	-2.4	0.04225222
hsa-miR-599	8.66	10.01	-1.16	0.5314607
hsa-miR-664b-5p	5.45	5.32	1.02	0.87722093
Disease control vs healthy control
Probe name	Disease control	Control	Disease control vs healthy control	P-value of: Disease control vs healthy control
hsa-miR-146b-3p	8.4	15.41	-1.83	0.04495602
hsa-miR-221-5p	11.84	18.22	-1.54	0.00787352
hsa-miR-4532	21.09	18.09	1.17	0.68054599
hsa-miR-599	10.01	18.5	-1.85	0.03426054
hsa-miR-664b-5p	5.32	9.86	-1.85	0.0062102

IgA: Immunoglobulin A.

The pathways enriched for the predicted targets of the 9-miRNA panel intersect several key processes implicated in IgAN pathogenesis. Multiple targets fall within mucosal immune and B-cell pathways, including cytokine and Toll-like receptor signalling, which are central to the generation of galactose-deficient IgA1 and anti-Gd-IgA1 autoantibodies at mucosal sites. Additional targets participate in nuclear factor-kappa B and Janus kinase/signal transducer and activator of transcription signaling pathways and chemokine networks that can drive mesangial inflammation and proliferation, linking upstream immune dysregulation to intrarenal injury. Enrichment of complement or coagulation cascades, extracellular matrix organisation, and transforming growth factor-beta-related signalling is consistent with the established roles of complement activation, glomerular damage, and progressive fibrosis in IgAN. These pathway-level observations, therefore, support the mechanistic plausibility of the 9-miRNA signature. The 9 miRNAs shortlisted from the above analysis were further analysed for their experimentally validated mRNA targets from miRTarBase. Pathways regulated by each miRNA were tabulated based on the mRNA targets (Figure 5).

Figure 5  Shows the network of nine microRNAs (yellow), pathways enriched (in boxes), experimentally targeted messenger RNA (purple oval), and disease association (in pink).

The 5 published miRNA biomarkers as observed from our previous study[13], which could, in combination, successfully differentiate healthy control IgAN cases with an AUC > 0.9, were studied within 3 comparisons (IgAN vs HC, DC vs HC, and IgAN vs DC). It was observed that the marker “hsa-miR-4532” is specific to IgAN, as it was found to be downregulated in IgAN and upregulated in DC. While the other 4 markers were found with the same direction of regulation within DC as well (downregulated in disease control samples) (Table 12)[33-43].

Table 12 Biological relevance and disease associations of the 9 candidate microRNAs¹.

miRNA	Ref.	Sample/context reported	Biological relevance/disease association
hsa-miR-4532	Seo et al[33], 2023; Kim et al[34], 2019; Liu et al[35], 2022	Urinary exosomes in diabetic kidney disease; urinary exosomes in kidney transplant recipients	Reported as a urinary exosomal biomarker in diabetic kidney disease, with significantly lower expression in biopsy-proven DKD compared with healthy controls, suggesting an association with chronic kidney injury; also part of a three-miRNA urinary exosomal signature (miR-21-5p, miR-31-5p, miR-4532) that discriminates acute rejection from stable graft function in kidney transplant recipients, indicating relevance to immune-mediated allograft injury; exosomal miR-4532 has been shown to promote endothelial cell injury via SP1 and NF-κB p65 activation, linking it to inflammatory and endothelial pathways relevant to glomerular and vascular damage
hsa-miR-4488	Zhong et al[36], 2021	Plasma-derived exosomes in dermatomyositis-associated interstitial lung disease	Identified among differentially expressed exosomal miRNAs in patients with dermatomyositis-associated interstitial lung disease, implicating roles in systemic autoimmunity and chronic inflammation; although kidney-specific data are not yet available, involvement in autoimmune and inflammatory settings supports potential relevance to immune-mediated glomerular injury
hsa-miR-3158-3p	Gupta et al[37], 2021; Gupta et al[38], 2026	Plasma in cerebral malaria; functional in immune-signaling assays	Elevated plasma miR-3158-3p levels correlate with MRI brain injury and poor outcome in cerebral malaria, a condition characterized by endothelial dysfunction and intense immune activation; functional studies show that miR-3158-3p overexpression downregulates NF-κB expression and modulates immune-related pathways, indicating a role in cytokine and innate immune signaling; no kidney-specific data are currently available, but the link to NF-κB and systemic inflammation is mechanistically compatible with IgAN-related immune activation
hsa-miR-151b	Király et al[39], 2024	Renal cell carcinoma tissue (hsa-miR-15b-5p and other family members studied)	Members of the miR-15b family (e.g., hsa-miR-15b-5p) are significantly downregulated in renal cell carcinoma compared with adjacent normal kidney and correlate inversely with tumor grade, suggesting roles in vascular/angiogenic and extracellular matrix pathways in kidney tissue; however, no kidney- or IgAN-specific functional data are currently available for hsa-miR-151b itself; no kidney/IgAN-specific functional data to date; identified here as a novel, hypothesis-generating diagnostic candidate
hsa-miR-3195	Zhou et al[40], 2021	Serum in Kallmann syndrome	Reported to target an aberrant PROK2 transcript and to modulate PROK2 function in vitro, with low serum expression associated with altered neuroendocrine phenotypes in Kallmann syndrome; while not kidney-specific, PROK2-related signaling can influence vascular and endocrine axes that may intersect with systemic immune and hemodynamic regulation; no kidney/IgAN-specific functional data to date; included as a novel biomarker candidate requiring further mechanistic work
hsa-miR-1289	Srivastava et al[41], 2023	In peripheral blood from patients diagnosed with COVID-19 infection	Cytosolic sulphonation of small molecules; amplification of signal from kinetochores; mitotic spindle checkpoint; no kidney/IgAN-specific functional data to date; identified here as a novel, hypothesis-generating diagnostic candidate
hsa-miR-20a-5p/hsa-miR-20b-5p	Donderski et al[42], 2022	Various kidney and CKD cohorts (miR-20a-5p family often profibrotic/profibrogenic in CKD panels)	miR-20 family members are frequently included among profibrogenic miRNA panels evaluated in CKD, where deregulation correlates with eGFR decline and proteinuria, and they are predicted to target components of TGF-β, cell-cycle, and apoptosis pathways associated with renal fibrosis; although direct IgAN-specific data are limited, their involvement in fibrotic and inflammatory signaling provides a plausible link to chronic glomerular/tubulointerstitial injury
hsa-miR-32-5p	Donderski et al[42], 2022	CKD-related miRNA panels (non-IgAN)	Included in several profiling studies as part of deregulated miRNA sets in chronic kidney disease and proteinuric states, with predicted targets in apoptosis, cell proliferation, and inflammatory pathways; no direct IgAN-specific functional studies, but pathway predictions intersect with NF-κB and TGF-β signaling implicated in glomerulosclerosis and tubulointerstitial fibrosis
hsa-miR-525-3p	Marques et al[43], 2011	In the peripheral circulation of hypertensive kidney disease	RHOF GTPase cycle; signal transduction; RHO GTPase cycle

¹Functional annotations are based on published literature and predicted or experimentally validated targets, and are intended as preliminary, hypothesis-generating information rather than definitive mechanistic assignments for immunoglobulin A nephropathy.

miRNA: MicroRNA; NF-κB: Nuclear factor kappa-light-chain-enhancer of activated B cells; DKD: Diabetic kidney disease; IgAN: Immunoglobulin A nephropathy; PROK2: Prokineticin 2; COVID-19: Coronavirus disease 2019; TGF-β: Transforming growth factor-β; CKD: Chronic kidney disease; RHOF GTPase: Ras homolog family member F.

DISCUSSION

IgAN is the most common cause of primary glomerulonephritis in the world. Despite this, no validated non-invasive biomarkers are currently available for its diagnosis[1]. Urinary exosomes provide a stable reservoir of miRNAs, making urine an attractive matrix for kidney biomarker discovery. Using NanoString technology, we profiled urinary exosomal miRNAs and identified multiple differentially expressed species in IgAN compared with healthy and DC. A panel of nine candidate miRNAs, as well as an alternative two-miRNA combination, discriminated IgAN from both control groups with AUCs exceeding 0.9 and 0.8, respectively. Yoon et al[44] studied the significance of urinary exosomal miRNA in IgAN, but they examined 12 miRNAs only based on previous publications, unlike this study, where a panel of 798 miRNAs was used to identify those miRNAs specific for IgAN. Another study by Li et al[45] looked at just 2 miRNAs in IgAN cases vs HC alone. DC were not included in their study.

Strengths of the study

This work is, to our knowledge, the first to compare healthy and diseased controls to ascertain the urinary exosomal miRNA signature in patients with IgAN. Additionally, it is the first study to be done exclusively with Indian participants. Also, this is the first study to use nanostring technology to identify the biomarkers for IgAN. Unlike microarray techniques used in a study by Min et al[46], Nanostring technology does not require hybridisation or amplification, making its results highly robust and precise. Furthermore, this study identifies nine miRNAs specifically associated with IgAN.

An important limitation of our validation strategy is that publicly available datasets with urinary exosomal miRNA profiles in IgAN are not yet available. Instead, we validated our candidate miRNAs using kidney biopsy tissue (GSE141344) and urinary sediment (GSE64306) datasets. These matrices differ biologically and technically from urinary exosomes: Kidney tissue captures intrarenal miRNA expression across glomerular and tubulointerstitial compartments, while urinary sediment includes a heterogeneous mixture of whole cells, cellular debris, and non-exosomal extracellular vesicles, whereas urinary exosomes represent a more defined, vesicle-encapsulated fraction shed from nephron and urinary tract epithelia. Differences in cellular composition, RNA content, and pre-analytical processing between tissue, sediment, and exosomal fractions are therefore likely to influence miRNA abundance and patterns, and may attenuate the direct biological comparability of our findings. Consequently, the external validation presented here should be interpreted as supportive of the diagnostic potential of these miRNAs across complementary biological compartments, rather than as a direct replication of a urinary exosomal signature. Future studies using independent cohorts with urinary exosomal miRNA profiling in IgAN will be essential to confirm or refine this biomarker panel.

Second, the IgAN cohort in this study was clinically heterogeneous, with a wide range of eGFR and proteinuria values at the time of biopsy (mean eGFR approximately 45 mL/minute/1.73 m² and mean proteinuria 2.69 g/day). Although Oxford MEST scores were recorded (M1: 98%, E1: 46%, S1: 26%, T1/T2: 50%), the sample size and event distribution did not allow adequately powered subgroup analyses to test whether individual urinary exosomal miRNAs or the 9-miRNA panel differed across histological classes or levels of kidney function. As a result, we cannot determine from this dataset whether the identified miRNA signature is stage-dependent or stable across the spectrum of IgAN severity, and this remains an important focus for future validation studies in larger, longitudinal cohorts.

CONCLUSION

Our research involving an Indian cohort found notable alterations in urinary exosomal miRNA profiles among individuals with IgAN, as opposed to both healthy and disease control groups. Such results emphasize the promise of miRNAs as non-invasive indicators for the diagnosis of IgAN.