{
  "abstract": "Introduction Renal involvement in type 2 diabetes (T2DM) can be due to diabetes (diabetic kidney disease, DKD) or other causes (non-DKD, NDKD) or both (mixed kidney disease). Available clinical and laboratory parameters have limitations in predicting a diagnosis (gold standard renal biopsy). Long non-coding RNAs (lncRNAs) evaluated in preclinical models but unexplored in biopsy-proven kidney disease in T2DM. We aimed to determine whether there is differential expression of lncRNAs in DKD (compared with NDKD).Research design and methods lncRNAs preselected through database search for evaluation in humans.Discovery cohort: Preselected lncRNAs () evaluated in three components of urine (urinary cell, urinary exosome, and cell-free urine) from biopsy-proven DKD, NDKD, T2DM without kidney disease and healthy subjects (n=40/group). lncRNAs found consistently significant in all components were checked in kidney tissue. Receiver operating characteristic curves were performed to evaluate diagnostic performance.Validation cohort: Best performing lncRNA (in discovery cohort) evaluated in independent cohort.Clinical utility: The utility of identified lncRNAs was further assessed for clinical decision-making.Results Discovery cohort: Level of MALAT1 and PVT1 differed in all urinary components of DKD and elevated in kidney biopsy tissue. MALAT1 showed the most consistent results. Urinary cell-derived MALAT1 showed the most consistent results (ΔCt <8.3, sensitivity 90%, specificity 89.6% OR 53.9, p<0.0001) to differentiate DKD from NDKDValidation cohort: Urinary cell-derived MALAT1 showed sensitivity (90%) and specificity (88.5%).Clinical utility: Addition of MALAT1 to currently existing clinical/biochemical discriminators of DKD from NDKD helps improve clinical decision making, with a net reclassification improvement (NRI) of 0.53, 64% of DKD cases were correctly reclassified to a higher probability of disease (NRI+ = 0.280), and 62.5% of NDKD controls were correctly reclassified to a lower probability of disease (NRI− = 0.250).Conclusions Urinary cell-derived MALAT1 improves the ability to differentiate DKD from NDKD over and above currently used clinical and biochemical parameters.",
  "authors": [
    {
      "affiliations": [
        "Department of Endocrinology and Metabolism, Institute of Postgraduate Medical Education and Research, Kolkata, India"
      ],
      "name": "Madhurima Basu"
    },
    {
      "affiliations": [
        "Department of Endocrinology and Metabolism, Institute of Postgraduate Medical Education and Research, Kolkata, India"
      ],
      "name": "Subhasis Neogi"
    },
    {
      "affiliations": [
        "Department of Endocrinology and Metabolism, Institute of Postgraduate Medical Education and Research, Kolkata, India"
      ],
      "name": "Ranu Pal"
    },
    {
      "affiliations": [
        "Department of Endocrinology and Metabolism, Institute of Postgraduate Medical Education and Research, Kolkata, India"
      ],
      "name": "Pradip Mukhopadhyay"
    },
    {
      "affiliations": [
        "Department of Nephrology, Institute of Postgraduate Medical Education and Research, Kolkata, India"
      ],
      "name": "Arpita Ray Chaudhury"
    },
    {
      "affiliations": [
        "Department of Endocrinology and Metabolism, Institute of Postgraduate Medical Education and Research, Kolkata, India"
      ],
      "name": "Nitai P Bhattacharyya"
    },
    {
      "affiliations": [
        "Department of Endocrinology and Metabolism, Institute of Postgraduate Medical Education and Research, Kolkata, India"
      ],
      "name": "Sujoy Ghosh"
    }
  ],
  "full_text": "WHAT IS ALREADY KNOWN ON THIS TOPIC Renal involvement in type 2 diabetes can be due to diabetes known as diabetic kidney disease (DKD) or can be due to causes other than diabetes, that is, non-DKD (NDKD) and may be an overlap of both (mixed disease). Renal biopsy remains the gold standard for differentiating DKD from NDKD. Currently used clinical markers have limitations in helping preselect subjects who might benefit from renal biopsy. NDKD patients may benefit from specific therapies, which could change clinical outcomes. Among various molecular signatures, non-coding RNAs including long non-coding RNAs (lncRNAs) are emerging as promising disease biomarkers. Changes in the expression of lncRNAs have been reported in animal and cell models of diabetes and kidney disease; however, human data, especially from biopsy-proven DKD and NDKD, remain limited.WHAT THIS STUDY ADDS In this study, urinary cell-derived MALAT1, a lncRNA (ΔCt <8.3), was able to differentiate DKD from NDKD, with 90% sensitivity and 89.6% specificity, and could predict DKD with an OR (OR 53.9, p<0.0001). In an independent validation cohort, it demonstrated a sensitivity of 90% and a specificity of 88.5%.Addition of MALAT1 to currently used parameters (clinical/biochemical discriminators) helps in the differentiation of DKD from NDKD and can help improve clinical decision making more accurately, with a net reclassification improvement (NRI) of 0.53, 64% of DKD cases were correctly reclassified to a higher probability of disease (NRI+ = 0.280), and 62.5% of NDKD controls were correctly reclassified to a lower probability of disease (NRI− = 0.250).In the exploratory analysis of 4-year follow-up data, baseline expression of MALAT1 was significantly elevated in those who developed composite renal outcomes compared with those who did not.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Urinary cell-derived MALAT1 improves the ability to differentiate DKD from NDKD over and above currently used clinical/biochemical parameters.Introduction Renal involvement in type 2 diabetes (T2DM) occurs because of diabetes itself, known as diabetic kidney disease (DKD), or from causes other than diabetes, such as IgA nephropathy, glomerulonephritis, hypertensive nephrosclerosis, and other conditions, collectively referred to as non-DKD (NDKD), or an overlap of both DKD and NDKD (mixed kidney disease). Previous studies suggest that up to one-third of all T2DM subjects have either NDKD or mixed kidney disease. Specific treatment options are often available for various forms of NDKD, which could alter renal outcomes. 1 2 NDKD subjects have better renal outcomes when diagnosed and appropriately treated with specific therapies.3 Renal biopsy remains the gold standard and only reliable test for accurately differentiating DKD from NDKD. Performing renal biopsy in T2DM subjects with renal involvement is impractical and the decision to undertake a biopsy is based on institutional policy, nephrologist decision, and certain clinical and laboratory parameters.1 Findings from clinical studies1 and meta-analysis4 suggest currently used clinical predictors to discriminate DKD from NDKD are suboptimal for decision making at the individual level to decide undertaking a renal biopsy. Given the high prevalence of diabetes and renal involvement in T2DM, limited availability of resources, and the invasive nature and risks associated with renal biopsy, there is an urgent need to identify non-invasive biomarker/s to help preselect subjects (especially DKD) who do not require biopsy, allowing this procedure to be reserved for patients with suspected NDKD, where histological insights could guide targeted treatment.To address these diagnostic challenges, non-coding RNAs (ncRNAs) have been explored as non-invasive molecular signatures. Various types of ncRNA, long ncRNAs (lncRNA) represent a widely prevalent group and are typically longer than 200 nucleotides.5 6 LncRNAs are highly tissue specific, they are expressed in diverse tissues and are often different in various diseases.6 Differential expression of lncRNAs has been implicated in the pathogenesis of many diseases, including diabetes and kidney disorders, as supported by literature.6 7 Most studies exploring the role of lncRNAs in DKD have been conducted using animal or cell models. These studies have demonstrated that lncRNAs can modulate various biological processes such as apoptosis, autophagy, and mitochondrial dysfunction in podocytes, as well as promote renal tubular epithelial cell injury and interstitial fibrosis.8 9 However human data regarding lncRNA in diabetes and kidney disease is limited. Few clinical studies are available, but the major drawback of such studies is that, subjects were presumed to have a diagnosis of DKD (not biopsy proven) and did not differentiate or exclude cases with NDKD.10 Although lncRNAs are stable in various biological fluids but for detection of kidney-related abnormalities urine serves as an ideal biofluid as it is easy to collect and likely to reflect pathobiological events related to renal dysfunction.11 lncRNAs in urine can be found in urinary cell pellet, cell-free urine and urinary exosomes (uEVs). Among these components, the urine cell pellet contains multiple cell types originating from the kidney, including podocytes, tubular epithelial cells, immune cells, and stem or progenitor cells.12 13 While RNAs in cell-free urine are heterogeneous in nature, they may originate both from the bloodstream and cells of the urogenital system, primarily due to processes such as apoptosis and necrosis.14 uEVs data demonstrated that uEVs contain proteins and RNAs characteristic of various kidney cell types, including podocytes, proximal and distal tubular epithelial cells, and collecting duct cells. Moreover, studies have shown that plasma EV are unable to cross the glomerular barrier in health and disease, suggesting that most uEVs arise from the urogenital system (predominantly from the kidney).14 15 However, most existing studies have focused on lncRNAs derived from peripheral blood mononuclear cells (PBMCs), plasma, or serum, which may not accurately reflect kidney-specific pathology. Additionally, studies analyzing urinary lncRNAs have typically considered only a single source, either urinary cells, cell-free urine, or uEVs. A comprehensive profiling of lncRNAs across all the sources of lncRNA in urine is lacking.In this context, we undertook this study to identify lncRNA(s) that are different in subjects with biopsy-proven DKD compared with NDKD, and to evaluate their potential utility in differentiating DKD from NDKD.The overarching objective of the study was to comprehensively evaluate urinary lncRNA expression profiles (preselected from database search) across all three components of urine, which include urinary cell pellet, cell-free urine, and uEV in patients with biopsy-proven DKD and NDKD. These profiles were compared with appropriate control groups, including individuals with T2DM without kidney disease as well as healthy non-diabetic (normoglycemic) controls, in the discovery cohort. Statistically significant differentially expressed lncRNAs across all three urinary components were further tested in kidney tissue to confirm their renal origin. The most consistent and best-performing lncRNA(s) identified in the discovery cohort were subsequently validated in an independent cohort of patients with biopsy-proven DKD and NDKD to establish their diagnostic relevance. The addition of any possible non-invasive marker (lncRNA/s) could improve the ability to diagnose DKD and NDKD beyond the currently used clinical and biochemical parameters.Methodology Selection of long non-coding RNAs LncRNAs were selected through a comprehensive in-silico approach using the lncRNA Disease V.2.0 database 16, an updated and freely accessible resource that catalogs the association between lncRNAs and a wide range of diseases. This database integrates findings from multiple studies that report strong correlations between lncRNAs and disease pathogenesis, supported by both experimental and computational evidence. Keyword searches were performed using terms including diabetes mellitus, T2DM, hyperglycemia, kidney disease, and diabetic nephropathy, to identify relevant lncRNAs. The initial selection was further refined to include only those lncRNAs with established roles in key pathological features of diabetes and kidney disease, as demonstrated in animal models and cell culture systems. Additionally, to ensure data quality and minimize bias, only lncRNAs that exhibited acceptable amplification levels (Ct <30) and were detectable in more than 40% of study subjects were retained for further analysis.17 Based on these criteria, 12 lncRNAs, NEAT1, GAS5, MALAT1, HOTAIR, PVT1, CASC2, XIST, OIP5, MIAT, MEG3, CDKN2B-AS1, and H19 were selected for this study. This targeted, evidence-based selection approach enabled us to focus on hypothesis-driven, functionally relevant lncRNAs while avoiding the need for costly techniques like sequencing. By leveraging existing knowledge to guide investigation, the precision and efficiency of the study were enhanced. (figure 1)Figure 1Selection of lncRNA in-silico analysis. lncRNA, long non-coding RNA.Study design, participant selection, and lncRNA profiling This was a single-center study, approved by the institutional ethics committee. Written informed consent was obtained. The overall study protocol is described in figure 2.Figure 2Schematic overview of the study(Baseline and follow-up)design. ACR,albumin-to-creatinine ratio; DKD, diabetic kidney disease; eGFR, estimated glomerular filtration rate; ESRD, end-stage renal disease; lncRNA, long non-coding RNA; NDKD, non-diabetic kidney disease; NRI, net reclassification improvement; OPD, outpatient department; ROC, receiver operating characteristic; T2DM, type 2 diabetes.Discovery cohort For the discovery cohort, subjects with DKD or NDKD were included from a large, ongoing renal biopsy cohort. The biopsy cohort was established to comprise individuals with biopsy-confirmed diabetes and kidney disease. Clinical and follow-up data for a subset of this biopsy-confirmed cohort have been already published. 1 3 To summarize the biopsy-cohort, it was established by recruiting consecutive subjects aged >18 years with known T2DM and renal involvement from the outpatient clinic of the Department of Endocrinology of the Institute of Postgraduate Medical Education & Research (Kolkata, India). Participants were included based on the following criteria: estimated glomerular filtration rate (eGFR) between 30 and 60 mL/min/1.73 m² calculated by CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration equation, 2012) and/or spot urine albumin-to-creatinine ratio(uACR) >300 mg/g on two occasions, 3 months apart, following adequate glycemic and blood pressure control. Subjects with eGFR <30 mL/min/1.73 m² were excluded to avoid inclusion of advanced/burnt-out disease, and those with eGFR >60 mL/min/1.73 m² were excluded due to ethical concerns regarding renal biopsy. Those who had no contraindications for biopsy underwent ultrasound-guided renal biopsy using an automated biopsy gun performed by two trained nephrologists. Renal biopsy specimens were processed for light microscopy, immunofluorescence microscopy, and electron microscopy as per standard protocol. Histopathological classification was performed independently by two expert renal pathologists following International Society of Nephrology and Renal Pathology Society guidelines.18 Based on pathological findings, subjects were grouped into DKD, NDKD and mixed kidney disease in the cohort. For this study, the mixed kidney disease group was excluded to avoid confounding effects, and only subjects with DKD and NDKD were included to ensure a clear analysis.For the T2DM without kidney disease group, participants attending the diabetes outpatient department of the institute who have been diagnosed with T2DM with normal kidney function (uACR<30 mg/g, eGFR >90 mL/min/1.73 m²) were included. In addition, normoglycemic individuals who had fasting blood sugar <100 mg/dL, postprandial blood sugar <140 mg/dL, and glycated hemoglobin <5.7 and exhibited no evidence of renal involvement (uACR <30 mg/g, eGFR >90 mL/min/1.73 m2) were enrolled as a healthy control group. Clinical, demographic, and biochemical parameters were recorded for all four groups, for example, DKD, NDKD, T2DM without kidney disease and healthy controls.Long non-coding RNA expression profile in all components of urine 50 mL of urine samples were collected from subjects across all four groups, that is, biopsy-proven DKD and NDKD along with appropriate control groups including subjects with T2DM without kidney disease as well as healthy controls. Total RNA was extracted from urinary cells isolated from 30 mL of urine following the manufacturer’s instructions (Zymo Research, Cat. No. R1039, USA). From 2 mL of urine supernatant obtained after centrifugation at 3000 RPM, cell-free RNA was extracted using the TRIzol method according to the laboratory’s standard protocol (RNAiso Plus, Takara, Cat. No. 9190). Extracellular vesicles (exosomes) were isolated from urine using total exosome Isolation Reagent (Thermo Fisher) and characterized by nanoparticle tracking analysis, with further validation by atomic force microscopy ( online supplemental data). RNA from exosomes was isolated using the TRIzol method (RNAiso Plus, Takara). Quantity and quality of extracted RNA were assessed using an Eppendorf bio spectrometer by measuring absorbance at 260 nm and 280 nm. Samples with an A260/A280 ratio of approximately 2.0 were considered of good purity and included in the study. Reverse transcription was performed in a 20 µL final reaction volume with 100 ng of RNA using the PrimeScript 1st Strand cDNA Synthesis Kit (Takara Bio) as per the manufacturer’s protocol. Quantitative real-time PCR was used to evaluate expression levels of urinary preselected lncRNAs, using SYBR Green dye detection (Takara Bio) on an Applied Biosystems Quant Studio 3 system. Housekeeping gene 18S served as an endogenous control. Expression levels were calculated using the ΔCt method, where ΔCt is cycle threshold Ct of gene of interest-ΔCt of housekeeping genes, smaller ΔCt value indicates higher expression. Relative expression (fold change) was determined by the 2−ΔΔCt method. Relative expression (fold change) of target lncRNAs in a sample compared with control after normalization to housekeeping gene.SP110.1136/bmjdrc-2026-006123.supp1Supplementary data LncRNA expression in kidney tissue LncRNAs that were significantly altered in comprehensive urine analysis were subsequently evaluated in kidney tissue. Total RNA was isolated from renal biopsy tissue slides according to the manufacturer’s instructions (Zymo Research, Cat. No. R1007). Expression levels of these lncRNAs were then evaluated using real-time PCR, as described previously.Clinical utility and diagnostic potential of these lncRNAs were evaluated, and those found to be significant were further validated in an independent cohort.Validation cohort Consecutive subjects who were histopathologically diagnosed with either DKD or NDKD in the department of nephrology of the institute were included in the validation cohort. This independent group of DKD/NDKD subjects was recruited based on the same inclusion and exclusion criteria used for the discovery cohort. Histopathological classification followed the same criteria applied in the discovery cohort.LncRNA expression in validation cohort Most consistent and best-performing lncRNAs (found in the discovery cohort) were further evaluated in a validation cohort, and their clinical utility to differentiate DKD from NDKD was assessed in comparison with currently used clinical predictors.Follow-up study (exploratory) Biopsy-proven DKD subjects were followed up at 3 months interval and eGFR, proteinuria were documented for 4 years period from the date of biopsy. Follow-up study design is described in figure 2. Renal composite outcomes were evaluated in these subjects. Composite renal outcome was defined based on doubling of creatinine (decline in eGFR more than 57% confirmed by two separate estimations with a minimum of 30 days period in between), progression to ESRD (sustained eGFR <15% on two separate occasions at least 30 days apart with or without maintenance hemodialysis or renal transplant) and renal death (due to any complication of chronic kidney disease in patients who were not receiving renal replacement therapy even when indicated).3 Further analysis compared lncRNA expression profiles between subjects who developed composite renal outcomes versus those who did not.Sample size Discovery cohort Sample size was calculated using MedCalc software V.19.5.6 (free online trial). Assuming α error of 0.05 and power of 80%, corresponding to a null hypothesis area under the curve (AUC) of 0.5, and an expected AUC of 0.7 for discriminating performance. Under this assumption the minimum required number of subjects per group was estimated to be 40 for the discovery cohort.Validation cohort Sample size for validating comparison of proportions was calculated using a significance level α error of 0.05 and a power of 90%. Based on known difference in proportions (66% vs 20%), the calculated sample size was 46.Statistical analysis Data were tested for normality using the Kolmogorov-Smirnov test. Categorical variables were expressed as frequencies and percentages. For non-parametric comparisons, the Kruskal-Wallis test was applied for analyses involving more than two groups, and the Mann-Whitney U test was used for comparisons between two independent groups.Receiver operating characteristic (ROC) curve analysis was performed to determine an appropriate cut-off value for lncRNA (ΔCt) to differentiate DKD from NDKD. Binary logistic regression was conducted to identify predictors of DKD. The models were tested for covariates that differed significantly between DKD and NDKD (using a liberal p value threshold of <0.1) and that had biological plausibility for between-group differences. Covariates to predict DKD compared with NDKD were duration of diabetes since diagnosis, presence of retinopathy, presence of hematuria, proteinuria, and low-density lipoprotein (LDL). Correlation analyses were performed among covariates, and variance inflation factor (VIF) was calculated to assess multicollinearity. VIF value >5 was considered evidence for determination of multicollinearity affecting the regression model.In the validation cohort, the cut-off obtained from the discovery phase was applied to subjects with histopathologically confirmed DKD or NDKD in an independent cohort to evaluate its predictive value. Category-free net reclassification improvement (cfNRI) to evaluate the independent diagnostic value of identified lncRNA/s, we compared a base clinical model (comprising disease duration, kidney size, proteinuria, hematuria, and retinopathy) to an extended model incorporating identified lncRNA/s.For the follow-up study, differences between groups were assessed using the Mann-Whitney U test. χ² test was performed to compare proportions and risks for outcomes, and binary logistic regression was conducted to identify predictors of outcomes. Statistical Package for the Social Sciences (SPSS V.25) and R were used for data processing and analysis.Results Discovery cohort Description of the cohort Participants with histopathology-confirmed DKD and NDKD were consecutively enrolled from the biopsy cohort. In addition, participants with T2DM without kidney disease and healthy subjects were included as comparison groups. (n=40 samples in each group).Mean age of study population (n=160) was 51.34 ± 7.6 years, and 60% of participants were male. Comparison of clinical and biochemical parameters of healthy controls, T2DM without kidney disease, and biopsy-proven DKD and NDKD groups are summarized in table 1 and online supplemental table 1.SP210.1136/bmjdrc-2026-006123.supp2Supplementary data Table 1Baseline clinical and biochemical parameters of healthy, T2DM, DKD and NDKD (n=120)Healthy (n=40)T2DM with kidney disease (n=40)DKD (n=40)NDKD (n=40)P valueP value(DKD vs NDKD)Age (year)56 (52.13–56.15)49.5 (46.7–52.33)53 (48.32–55.33)52 (33–70)0.07 NSKnown duration of disease (months)051 (51.97–92.03)144 (113.15–169.85)43.5 (33.68–54.38)<0.0001<0.0001BMI (kg/m2)21.46 (20.98–21.87)21.69 (21.24–23.16)26.08 (24.57–27.27)23.4 (22.78–25.48)<0.00010.049HBA1C (mmol/mol)5.1 (4.89–5.17)7.33 (7.19–7.78)7.7 (7.75–8.82)6.34 (5.98–6.37)<0.0001<0.0001ACR (mg/g)6 (4.41–6.01)10.05 (9.89–16.33)578.42 (225.21–3048.25)567 (484.94–1576.55)<0.00010.69eGFR (CKD-EPI) (mL/min/1.73 m2)117 (99–120)88.2 (70–106.5)50 (39.3–65.9)60 (46.8–91.5)<0.0001NSPresence of nephrotic range proteinuria (%)0047.7545NSNSPresence of hematuria (%)0027.530NSNSPresence of retinopathy (%)00755NS<0.0001Absence of neuropathy (%)008033NS<0.0001Continuous data presented as median (IQR), p value was calculated by Mann-Whitney U test and categorical data as percentage (%) and p value was calculated by χ² test.ACR, albumin-to-creatinine ratio; BMI, body mass index; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration; DKD, diabetic kidney disease; eGFR, estimated glomerular filtration rate; HBA1C, glycated hemoglobin; NDKD, non-diabetic kidney disease; NS, not significant; T2DM, type 2 diabetes.lncRNA expression profile in all components of urine Urinary cell pellets Expressions of candidate lncRNAs were determined in urinary cell pellets from all four groups, results summarized in figure 3. Expression of PVT1 and MALAT1 were significantly uplifted in DKD as compared with both NDKD (p=0.0002, p<0.0001, respectively) and T2DM without kidney disease (p=0.0003, p=0.0003, respectively). In contrast, XIST expression was significantly downregulated in DKD compared with the T2DM without kidney disease group (p=0.04). Additionally, GAS5 expression was significantly upregulated in DKD compared with NDKD (p=0.02), with no significant difference found with the T2DM without kidney disease group.Figure 3Expression of urinary cell- pellet derived long non-coding RNA among healthy, T2DM without kidney disease, DKD and NDKD group. All data are presented as mean±SEM and were analyzed within two groups by Mann-Whitney U test and for four groups Kruskal-Wallis test (ns: p>0.05, *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001). DKD, diabetic kidney disease; NDKD, non-diabetic kidney disease; ns, not significant; T2DM, type 2 diabetes.Cell-free urine Expression of selected lncRNAs was evaluated in cell-free urine samples among all groups and summary of the results is shown in figure 4. Expression of PVT1 was significantly downregulated and MALAT1 was significantly upregulated in DKD compared with both NDKD (p=0.0001, p<0.0001, respectively) and T2DM without kidney disease (p=0.0008, p=0.04 respectively). Additionally, MIAT (p=0.03) and H19 (p=0.001) levels were altered in DKD compared with T2DM without kidney disease only. GAS5 expression was significantly downregulated in DKD compared with NDKD (p=0.001), however, no significant difference was observed with T2DM without kidney disease.Figure 4Expression of urinary cell-free long non-coding RNA among healthy, T2DM without kidney disease, DKD and NDKD group. All data are presented as mean±SEM and were analyzed within two groups by Mann-Whitney U test and for four groups Kruskal-Wallis test (ns: p>0.05, *p≤0.05, **p≤0.01,***p≤0.001, ****p≤0.0001). DKD, diabetic kidney disease; NDKD, non-diabetic kidney disease; ns, not significant; T2DM, type 2 diabetes.Urinary exosome Expression of prioritized lncRNAs was analyzed in uEV, summary result is described in figure 5. Notably, expression of MALAT1 was significantly upregulated in DKD compared with NDKD (p=0.0007) and T2DM without kidney disease (p=0.001). Conversely, expression of PVT1 was significantly downregulated in DKD relative to both NDKD (p=0.002) and T2DM without kidney disease (p=0.0002). Additionally, H19, and OIP5 showed altered expression in DKD compared with T2DM without kidney disease. However, no significant differences in GAS5 expression were observed in uEV among the groups.Figure 5Expression of urinary exosome containing long non-coding RNA among healthy, T2DM without kidney disease, DKD and NDKD group. All data are presented as mean±SEM and were analyzed within two groups by Mann-Whitney U test and for four groups Kruskal-Wallis test (ns: p>0.05, *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001). DKD, diabetic kidney disease; NDKD, non-diabetic kidney disease; ns, not significant; T2DM, type 2 diabetes.Comparison of urinary lncRNA expression profile Among 12 lncRNAs analyzed, the level of MALAT1 and PVT1 expression was significantly different across all three components of urine. MALAT1 expression showed a consistent and promising pattern in all components of urine (cellular, cell-free, and exosome). lncRNA profiling results indicated that among all analyzed lncRNAs, MALAT1 exhibited a progressive increase in expression from healthy controls to T2DM without kidney disease group, and further from T2DM without kidney disease group to DKD with significantly higher levels in DKD compared with NDKD in all three components of urine. In contrast, PVT1 expression was significantly altered in DKD compared with NDKD and T2DM without kidney disease, but its results were inconsistent.Kidney tissue To confirm that the source of the significantly altered lncRNAs was the kidney itself, expression levels of MALAT1 and PVT1 were assessed in kidney tissue ( figure 6). Both lncRNAs, MALAT1 (p=0.004) and PVT1 (p=0.004), were significantly upregulated in DKD compared with NDKD which confirms the renal origin of these lncRNAs.Figure 6Represents expression long non-coding RNA in renal tissue of DKD and NDKD subjects. (All data are presented as mean±SEM and were analyzed within groups by Mann-Whitney U test. (ns: p>0.05, *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001). DKD, diabetic kidney disease; NDKD, non-diabetic kidney disease; ns, not significant.Determination of cut-off and clinical performance analysis in discovery cohort ROC curve analysis was performed for both MALAT1 and PVT1 in urinary cell pellet, cell-free urine and exosome for diagnosis of DKD as compared with NDKD. Summary result is shown in figure 7. Urinary cell pellet-derived MALAT1 expression, using a ΔCt value <8.3 as the diagnostic threshold for DKD, demonstrated the highest sensitivity (90%) and highest specificity (89.6%) in distinguishing DKD from NDKD (AUC 0.95, 95% CI 0.9 to 1, p<0.0001).Figure 7ROC curve analysis for identifying DKD compared with NDKD. AUC, area under the curve; DKD, diabetic kidney disease; NDKD, non-diabetic kidney disease; ROC, receiver operating characteristic.To determine predictors for histopathological diagnosis of DKD, binary logistic regression analysis was performed, considering diagnosis DKD compared with NDKD as the dependent variable and duration of diabetes since diagnosis, presence of retinopathy, presence of hematuria, proteinuria, and LDL as covariates. Results indicated MALAT1 may serve as a predictor for diagnosis of DKD (OR 53.9, (95% CI 3.7 to 774.1) p<0.0001).Validation cohort Total 62 subjects were included in an independent validation cohort among them 37 subjects had biopsy-proven DKD and 25 had NDKD. Summary of clinical and baseline characteristics of the validation cohort is presented in table 2.Table 2Clinical and biochemical parameters in validation cohort (n=62)DKD (n=37)NDKD (n=25)P valueAge (year)49 (43.8–52.3)53 (46.29–59.18)0.2Known duration of disease (months)126 (90.4–150)43 (20.9–54.03)<0.001BMI (kg/m2)25.19 (22.83–27.5)18.49 (16–33.5)NSCholesterol (mg/dL)165 (119–222.6)247 (85–503.1)0.83Triglyceride (mg/dL)195 (129.4–258.22)280.6 (137–315)0.91HDL (mg/dL)45 (35.4–54.18)45.8 (18–133.3)0.95LDL (mg/dL)83.7 (41–134)93 (41–144.3)0.79Albumin (gm/dL)3.5 (3.1–3.6)2.9 (1.5–4.6)0.8HBA1C (mmol/mol)7.5 (6.6–10.5)5.7 (3.7–8.1)<0.01Urea (mg/dL)37.5 (32.8–44.4)29 (17–99.1)0.3Creatinine (mg/dL)1.44 (0.98–1.77)1.08 (0.8–2.4)0.19ACR (mg/g)538.42 (325–3548.25)467 (384.94–1680)0.7eGFR (CKD-EPI) (mL/min/1.73 m2)58.5 (36.5–82.3)94.8 (36–166)0.2624 hours protein (mg)3500 (2500–6670)3048 (1487–4488)NSFamily history of diabetes (%)61500.1Family history of kidney disease (%)38400.3Unequal kidney (%)1100.5High C3 level (%)00NSHigh C4 level (%)16160.6ANA positive (%)100NSANCA positive00NSPR3/MPO/Anti-GBM positive00NSPresence of nephrotic range proteinuria (%)47.7545NSPresence of hematuria (%)1140NSUrinary cast (%)1512NSPresence of hypertension (%)88500.1Presence of retinopathy (%)725<0.001Absence of neuropathy (%)9025<0.001Continuous data presented as median (IQR), p value was calculated by Mann-Whitney U test and categorical data as percentage (%) and p value was calculated by χ² test.ACR, albumin-to-creatinine ratio; ANA, antinuclear antibody; ANCA, anti-neutrophil cytoplasmic antibodies; BMI, body mass index; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration; DKD, diabetic kidney disease; eGFR, estimated glomerular filtration rate; GBM, glomerular basement membrane; HBA1C, glycated hemoglobin; HDL, high-density lipoprotein; LDL, low-density lipoprotein; MPO, myeloperoxidase; NDKD, non-diabetic kidney disease; NS, not significant; PR3, proteinase 3.In this validation cohort, urinary cell pellet-derived MALAT1 (using cut-off from discovery cohort) was found to effectively distinguish DKD from NDKD, with a sensitivity of 88.5% and a specificity of 90%, implying that MALAT1 is a useful marker for differentiating DKD from NDKD.Utility of MALAT1 in clinical decision-making To evaluate the utility of MALAT1 in improving clinical decision-making in differentiating DKD from NDKD over and above currently used clinical and biochemical parameters (comprising disease duration, kidney size, proteinuria, hematuria, and retinopathy) was compared with an extended model incorporating MALAT1. cfNRI for the addition of MALAT1 was 0.530 (95% CI 0.104 to 0.910), indicating significant enhancement in diagnostic accuracy. Specifically, 64% of DKD cases were correctly reclassified to a higher probability of disease (NRI+ = 0.280), and 62.5% of NDKD controls were correctly reclassified to a lower probability of disease (NRI− = 0.250). These findings suggest that MALAT1 provides significant incremental utility in correctly identifying patient diagnostic status compared with standard clinical parameters alone.Association of MALAT1 with histopathological parameters To assess the potential association between urinary cell-pellet derived MALAT1 expression and histopathological parameters Kendall’s tau-b analysis was performed. Results revealed that MALAT1 expression was also positively correlated with histopathological parameters including mesangial expansion (r=0.424, p<0.001), Kimmelstiel-Wilson lesions (r=0.219, p=0.001), arteriolar hyalinosis (r=0.428, p<0.001), thickening of both tubular and glomerular basement membranes (r=0.360, p=0.002 and r=0.466, p<0.001, respectively) and negatively correlated with IgA (0.291, p=0.01), mild mesangial hypercellularity (r=0.304, p=0.009).Association of MALAT1 with adverse renal outcome (follow-up study) DKD subjects were followed up for 4 years to look for composite renal outcome. Among 51 DKD subjects who had completed the follow-up, 32 of them had developed composite renal outcome where 19 of them had no renal outcome. Baseline expression of MALAT1 was compared, where expression of MALAT1 was significantly elevated in those who developed composite renal outcomes compared with those who did not ( figure 8). Furthermore, logistic regression analysis, with presence and absence of composite renal outcome as dependent variable and MALAT1 expression and duration of diabetes as covariates, indicated that baseline MALAT1 expression can predict renal outcomes in DKD subjects (OR=7.3, 95% CI 1.2 to 41.66, p=0.025).Figure 8Expression of urinary cell pellet derived MALAT1 among DKD subjects with and without renal outcome. All data are presented as mean±SEM and were analyzed within two groups by Mann-Whitney U test (ns: p>0.05, *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001). DKD, diabetic kidney disease.Discussion This study aimed to evaluate the profile of lncRNAs expressed in urine. 12 lncRNAs were pre-selected through database search. Among those evaluated lncRNAs, GAS5 showed altered expression in cell-free and cellular urine, but not in uEV. Although PVT1 expression was significantly altered across all urinary components, the result lacked consistency. Notably, only lncRNA MALAT1 demonstrated consistent expression patterns in cell-free urine, urinary cell pellet, uEV, and kidney tissue. Thus, MALAT1 emerged as a potential marker of DKD effectively distinguishing it from both T2DM without kidney disease and NDKD. The renal origin of MALAT1 was further confirmed by its expression in kidney tissue. Established cut-off from the discovery cohort was validated in an independent cohort underscoring the robustness of our findings. Diagnostic performance of MALAT1 in clinical decision making demonstrated that it outperforms current models in distinguishing DKD from NDKD. Further exploratory analysis of 4 years follow-up data indicates that baseline MALAT1 expression is upregulated in DKD subjects who developed renal outcomes compared with those who did not.All preselected lncRNAs have previously been reported to be significantly associated with DKD in vitro cell culture19 or animal models,20 implicating their involvement in DKD pathogenesis.21 22 However, findings from those preclinical studies did not show concordant expression in biopsy-proven DKD, the gold standard test.Previous studies have reported altered urinary MALAT1 expression in type 1 diabetes and kidney disease23 and MALAT1 expression in blood has been shown to be increased in T2DM without kidney disease and DKD relative to healthy control.24 However, these studies defined DKD solely by uACR and eGFR, lacked validation in biopsy confirmed DKD and did not incorporate established clinical predictors for renal biopsy, and were limited to comparisons between DKD and healthy controls only. To the best of our knowledge, no comprehensive study has evaluated MALAT1 expression across biopsy-proven DKD, NDKD, and control groups such as T2DM without kidney disease and healthy individuals.Moreover, expression of these lncRNAs has primarily been evaluated in PBMCs, plasma, or serum in previous studies. While few studies have explored lncRNAs in urine, data regarding comprehensive urinary profile analysis including all components (ie, cell-free urine, uEV, and urinary cell pellets) is lacking. RNA expression in circulating urine is relatively low, and RNA extraction from exosome is labor-intensive and time-consuming, limiting its use in clinical practice. In contrast, urinary cell pellets contain high levels of nucleic acids and are less influenced by humoral metabolic factors, making them practical as a non-invasive source of lncRNA.16 25 Several previous studies have employed advanced sophisticated techniques such as RNA sequencing26 to identify lncRNAs in an untargeted manner, though few have been conducted in human cohorts. Despite the use of high-cost advanced methodologies, the clinical translational potential of many identified lncRNAs remains limited. This is largely due to the lack of validation in adequately powered biopsy-proven DKD cohort, which is essential for establishing clinical relevance. In contrast, our approach involved systematic database mining, incorporating insights from those earlier studies establishing the role as a marker.Unlike previous studies that only reported differential expression of lncRNAs without in-depth statistical evaluation, our study integrates robust statistical approaches, including ROC, logistic regression, and NRI to establish the clinical relevance of MALAT1.We acknowledge that the use of NRI has been a subject of statistical debate, with critics citing its sensitivity to model misspecification and the potential for overestimating clinical utility when applied to the same data used for model training. However, we believe reliance on NRI in this study is both justified and robust for several reasons. First, to mitigate the risk of “optimism bias,” our NRI metrics were calculated exclusively on an independent validation cohort rather than the discovery set. This ensures that observed reclassification reflects the biomarker’s true generalizability rather than statistical noise. Second, NRI provides a more granular view of how MALAT1 “corrects” the diagnostic status of patients who were previously misclassified by currently used clinical and laboratory discriminators. Notably, MALAT1 improved the diagnosis of both DKD and NDKD, indicating its utility across the full disease spectrum.This study has several notable strengths; a key advantage is the use of biopsy-proven DKD and NDKD, the gold standard for diagnosis. The study also provides a comprehensive evaluation of lncRNA expression across all urinary components offering a more complete understanding of urinary lncRNA profiles. Among the candidates, MALAT1 emerged as a robust and consistent biomarker, demonstrating concordant expression supporting its renal origin. Furthermore, the focus on urinary cell pellets as a practical, non-invasive source of RNA enhances translational potential, and preliminary longitudinal data suggest a possible prognostic role of MALAT1 in predicting renal outcomes.This study is particularly relevant for about two-thirds of individuals with T2DM and renal involvement (DKD cases) where kidney biopsy could potentially be avoided and more importantly help preselect subjects who could potentially benefit from biopsy (NDKD). However, several limitations should be acknowledged. Given that this study is based on a single-center cohort, it should be considered a proof of concept. Additionally, while MALAT1 showed strong diagnostic performance, the study provides limited mechanistic insight into its role in DKD pathogenesis. Technical challenges related to low RNA yield and variability in urine-based assays, identified lncRNAs require extensive validation across multicenter, multinational, and multiethnic populations, as well as through high-throughput analyses, to enhance their translational potential before they can be reliably implemented in clinical settings. Even exploratory analyses further indicated an association of MALAT1 with adverse histopathological changes and poor renal outcomes in DKD, strengthening our observations; however, adequately powered longitudinal studies are still required.Summary and conclusion LncRNAs associated with diabetes and kidney disease were identified through database search were tested for expression in all three components of urine (cell pellet, cell-free RNA and in exosomes) of subjects (Healthy controls, T2DM without kidney disease, biopsy-proven DKD and NDKD). Level of MALAT1 was consistently found to be different (in discovery cohort) with best expression in urinary cell pellet (with ΔCt <8.3 identifying DKD with high sensitivity (90%) and specificity (89.6%) (OR 53.9). Expression of MALAT1 was checked in biopsy tissue to confirm its renal origin. These findings were confirmed in a validation cohort, where MALAT1 demonstrated robust sensitivity (88.5%) and specificity (90%). Incorporation of MALAT1 into the current diagnostic algorithm (including currently used clinical/laboratory parameters) significantly improved diagnostic accuracy, with NRI of 0.53, implying that MALAT1 adds discriminatory value beyond currently used markers and may aid in differentiating DKD from NDKD.SP310.1136/bmjdrc-2026-006123.supp3Supplementary data",
  "title": "Integrated multicompartment urinary long non-coding RNAs profiling (cellular, cell-free, and extracellular vesicle) for better differential diagnosis of biopsy-proven diabetic and non-diabetic kidney disease beyond conventional markers",
  "uid": "167a7a72-0e68-5ca1-86bf-440fa335f138"
}