Abstract
Nephrotic syndrome is the leading cause of pediatric chronic kidney disease, particularly the steroid-resistant subtype (SRNS). The current diagnostic paradigm relies on prolonged steroid trials or invasive biopsies, often delaying optimal treatment. This study systematically synthesizes evidence from cohort studies on non-invasive molecular biomarkers predicting SRNS at onset. We searched PubMed, Scopus, ScienceDirect, and MEDLINE Ultimate via EBSCO for cohort studies published between January 2015 and September 2025. The study protocol was registered with PROSPERO (CRD420251147997), and quality was assessed using the Newcastle-Ottawa Scale. Eleven studies comprising 718 study-level participant observations were included; as two studies originated from an overlapping cohort within a single research group, the number of unique children is likely lower. Candidate molecular signatures were identified across plasma proteins, cytokines, metabolites, urinary markers, transcriptomic markers, and a genetic/epigenetic model. Multi-marker panels showed high predictive utility, including a plasma protein panel (VDB+ADIPOQ+MMP2; AUC 0.780; P = 0.003) and a cytokine panel (IL-7+IL-9+MCP-1; AUC 0.846; sensitivity 64.3%; specificity 84.6%). Among individual markers, TGF-β2 mRNA showed the highest predictive accuracy (AUC 0.972; P < 0.001) in a small single cohort, followed by plasma glutamine in an age-combined model (AUC 0.810; OR 1.01), while reduced plasma malonate showed predictive value in children older than 3 years (AUC 0.84; OR 0.94). Conversely, urinary VDBP failed to distinguish SRNS from SSNS (P = 0.417), suggesting possible confounding by proteinuria severity. These exploratory findings suggest that several molecular biomarkers, particularly composite panels, may hold promise for future early risk stratification; however, given the small number and size of cohorts, participant overlap, and methodological heterogeneity, these results should be interpreted as hypothesis-generating rather than evidence of superiority of any single biomarker class, requiring validation in larger, independent, prospective cohorts before clinical application.
Keywords: biomarkers, nephrotic syndrome, pediatric, steroid-resistant
Introduction
Nephrotic syndrome stands as the most common chronic glomerular pathology in pediatrics. Globally, the incidence ranges from 2 to 17 cases per 100,000 children per year, though epidemiological data suggests substantial variability influenced by genetic background and ethnicity.1–3 Based on the response to standard corticosteroid therapy, the disease is dichotomized into steroid-sensitive nephrotic syndrome (SSNS) and the prognostically severe steroid-resistant nephrotic syndrome (SRNS).
Clinically, SRNS is defined as the absence of remission after four weeks of daily prednisone administration at a dosage of 2 mg/kg/day. While SRNS accounts for only 20% of total nephrotic syndrome cases, it poses a significant clinical burden. Histopathologically, SRNS is frequently correlated with focal segmental glomerulosclerosis (FSGS) and is a leading cause of renal progression in children.3 Pathophysiologically, SRNS often reflects underlying podocyte injury or genetic mutations affecting the glomerular filtration barrier, which in a subset of cases renders standard corticosteroid therapy ineffective.4 While SRNS is thought to be largely driven by immune dysregulation, SRNS is pathophysiologically more heterogeneous, encompassing both structural/genetic and immune-mediated mechanisms; this heterogeneity necessitates distinct predictive markers beyond standard clinical presentation.4,5 Data indicate that SRNS contributes to 10.4% of all pediatric chronic kidney disease (CKD) cases, with a rapid progression rate to end-stage renal disease (ESRD) occurring in 15.2% to 24.3% of patients.6 This high risk of renal failure underscores the urgent need for predictive tools to guide early intervention.7
The current diagnostic paradigm for differentiating SRNS from SSNS is suboptimal. Definitive classification often requires either an invasive renal biopsy, which carries procedural risks, or a wait-and-see approach involving steroid administration.8,9 The latter is particularly problematic, as it exposes children with SRNS to adverse effects of prolonged steroid use without therapeutic gain, while delaying potentially effective second-line immunosuppression.2,3,10 To address these limitations, recent research has pivoted toward identifying non-invasive molecular biomarkers capable of predicting steroid resistance at onset. In the era of precision medicine, one long-term aspiration is to move beyond a uniform steroid trial toward more individualized management.11 If adequately validated in future prospective studies, such biomarkers could eventually help guide earlier initiation of calcineurin inhibitors or genetic screening in selected high-risk children, potentially sparing them from unnecessary steroid toxicity.
However, a gap remains in the systematic synthesis of this evidence. Prior systematic reviews of biomarkers in nephrotic syndrome have often combined adult and pediatric populations or pooled case-control and cohort designs, limiting their ability to establish true predictive value specifically in children.12 This review addresses that gap by focusing exclusively on pediatric, pre-treatment cohort studies published over the last ten years (2015–2025). Cohort studies were prioritized in this review because their longitudinal pre-treatment sampling design allow true predictive value to be assessed, in contrast to case-control or cross-sectional designs that cannot distinguish whether a biomarker precedes or results from disease activity. In this review, a predictive biomarker is defined as one measured prior to treatment initiation that forecasts subsequent steroid responsiveness, as distinct from a prognostic biomarker, which forecasts long-term disease course irrespective of treatment. The research question was framed as: “in children with idiopathic nephrotic syndrome, do pre-treatment molecular biomarkers, compared with standard clinical assessment, predict steroid resistance?”. The objective is to identify and evaluate candidate molecular biomarkers; ranging from urinary and plasma proteomics to metabolomic, epigenetic, and cytokine signatures; for their predictive value for SRNS in children, as a step toward informing the eventual transition from empirical treatment to precision medicine.
Methods
This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement.13 The study protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) under the registration number of CRD420251147997.
Eligibility Criteria
Studies were selected based on the following inclusion criteria: 1) studies evaluating and comparing molecular biomarkers for predicting SRNS in pediatric populations; 2) studies reporting statistical analyses of biomarker performance such as sensitivity, specificity, odds ratios, or area under curve (AUC) of receiver operating characteristics (ROC); 3) full-text articles available in English; 4) studies published between January 1, 2015 and September 23, 2025; and 5) observational studies with a cohort design. Exclusion criteria included: 1) systematic reviews, meta-analyses, editorials, commentaries, or expert opinions; 2) animal or in vitro studies; and 3) conference abstracts without available full texts or unpublished data.
Search Strategy
A comprehensive search was performed across four electronic databases: PubMed, ScienceDirect, Scopus, and MEDLINE Ultimate via EBSCO. The search was limited to publications from the last ten years (2015–2025). The search strategy utilized a combination of Medical Subject Headings (MeSH) and free-text keywords, including terms related to “Biomarkers”, “Nephrotic Syndrome”, “Steroid Resistance”, and “Children”. Boolean operators (AND/OR) were adapted to the specific syntax requirements of each database. An initial search was conducted on September 23, 2025. Following peer review, the search was re-run in its entirety across all four databases on August 6, 2026, using the same search strategy and the same publication-date eligibility window (January 1, 2015 through September 23, 2025); the records, screening results, and included studies reported in this manuscript and in the PRISMA flow diagram (Figure 1) reflect this August 6, 2026 search. The complete search strings for each database are detailed in Supplementary Table 1.
Figure 1.
PRISMA flow chart of study search and selection.14
All identified records were imported into Rayyan AI for duplicate removal and efficient screening. Initial screening of titles and abstracts was performed independently by three reviewers (AHP, GAR, ROS) to identify potentially relevant studies. Subsequently, full-text articles were retrieved and assessed against the eligibility criteria. Any discrepancies during the screening process were resolved through discussion or consultation with two senior reviewers (AW and YP). In addition to the database search, backward citation searching was performed on the reference lists of all initially eligible full-text articles to identify additional studies not captured by the database search.
Data Extraction
Data extraction was performed using a standardized data collection form. Information extracted included: study characteristics (first author, year of publication, country, study design), population demographics (sample size, age distribution), details of the biomarker (type, biological specimen, measurement method), statistical analysis methods, and key outcomes regarding predictive performance. Data extraction was conducted independently by three reviewers (AHP, GAR, ROS) and validated by two senior reviewers (AW and YP). To mitigate conflict of interest, reviewers who were authors of any included study were recused from extracting data or assessing the quality of their own work. For studies employing a paired pre- and post-treatment design, only baseline (pre-treatment) data were extracted and analyzed, consistent with this review’s focus on predictive biomarkers measured prior to steroid initiation.
Risk of Bias
The methodological quality and risk of bias of the included cohort studies were assessed using the Newcastle-Ottawa Scale (NOS). Studies were evaluated based on three domains: selection of study groups, comparability of the groups, and ascertainment of the outcome. The assessment was performed independently by three reviewers (AHP, GAR, ROS). Disagreements were resolved through consensus or arbitration by senior reviewers (AW and YP). As in the data extraction phase, reviewers were excluded from assessing the quality of their own publications to ensure objectivity.
Results
Search Results and Study Selection
A comprehensive search of four electronic databases (PubMed, ScienceDirect, Scopus, and MEDLINE Ultimate via EBSCO) yielded a total of 882 records. After removing 450 duplicates, 432 titles and abstracts were screened. Of these, 32 full-text reports were sought for retrieval, of which 3 could not be retrieved, leaving 29 reports assessed for eligibility. Twenty-one reports were excluded (19 for wrong study design, 1 not published in English, and 1 for reporting a combined steroid-dependent/steroid-resistant outcome rather than an SRNS-specific outcome), resulting in 8 studies identified through database searching.
In addition, backward citation searching was performed on the reference lists of all initially eligible full-text articles, yielding 277 records. Of these, 25 were sought for full-text retrieval, of which one could not be retrieved, leaving 24 reports assessed for eligibility. 21 reports were excluded (9 for wrong study design, 3 as duplicates of studies already identified via database searching, and 9 for duplicating each other within the snowballed set), resulting in three additional eligible studies identified through citation searching.
In total, eleven cohort studies met the inclusion criteria and were included in this systematic review.15–25 This selection process is detailed in the PRISMA flow diagram14 (Figure 1).
Characteristics of Included Studies
The included studies were cohort studies published between 2015 and 2025, conducted across eight countries: the United States, India, China, Egypt, Indonesia, Poland, Italy, and Iran. Crucially, all studies utilized a longitudinal design in which biological samples were collected at disease presentation (pre-treatment/baseline) prior to the initiation of steroid therapy, allowing for the assessment of true predictive value (Table 1).
Table 1.
Characteristics and Data Extraction of Included Studies
| Author, Year (Country) | Study Design, Total Sample (SRNS/SSNS) | Sampling Time Point, Follow Up Duration | Specimen Type, Detection Method | Biomarkers | Main Findings | Sensitivity; Specificity; AUC (P-value) |
|---|---|---|---|---|---|---|
| Agrawal et al, 2020 (USA)24 | Cohort, 37a (13/24) | Pre- and post-treatmentb, 7 weeks | Blood, proteomic discovery and immunoblotting validation | Panel of three proteins: | ||
| VDB + ADIPOQ + MMP2 | Combination of three markers significantly predicts SRNS at onset. | NR; NR; 0.78 (0.003) | ||||
| Individual proteins: | ||||||
| COL6A3 | Lower in SRNS | NR; NR; NR (0.02) | ||||
| IGFBP2 | Lower in SRNS | NR; NR; NR (0.04) | ||||
| MMP2 | Lower in SRNS | NR; NR; NR (0.03) | ||||
| APOE | Lower in SRNS | NR; NR; NR (0.04) | ||||
| ADIPOQ | Lower in SRNS | NR; NR; NR (0.01) | ||||
| SHBG | Lower in SRNS | NR; NR; NR (0.03) | ||||
| EFEMP1 | Lower in SRNS | NR; NR; NR (0.01) | ||||
| ITIH4 | Lower in SRNS | NR; NR; NR (0.01) | ||||
| HPX | Higher in SRNS | NR; NR; NR (0) | ||||
| Plasma VDB | Higher in SRNS | NR; NR; NR (0.03) | ||||
| SERPINC1 | Higher in SRNS | NR; NR; NR (0.01) | ||||
| AZGP1 | Higher in SRNS | NR; NR; NR (0.03) | ||||
| FETUB | Higher in SRNS | NR; NR; NR (0.02) | ||||
| Agrawal et al, 2021 (USA)21 | Cohort, 40 (16/24) | Pre- and post-treatmentb, 7 weeks | Blood, multiplex bead-based immunoassay (Luminex) | Multivariate panel: | ||
| IL-7 + IL-9 + MCP-1 | Combination of three cytokines significantly predicts SRNS at onset. | NR; NR; 0.846 (<0.001) | ||||
| Individual cytokines: | ||||||
| IFN-γ | Higher in SRNS | NR; NR; 0.792 (<0.004) | ||||
| MIP-1⍺ | Higher in SRNS | NR; NR; 0.774 (0.007) | ||||
| IL-6 | Higher in SRNS | NR; NR; 0.740 (0.017) | ||||
| IL-1β | Higher in SRNS | NR; NR; 0.723 (0.027) | ||||
| TNF-⍺ | Higher in SRNS | NR; NR; 0.679 (0.075) | ||||
| IL-7 | Higher in SRNS | NR; NR; 0.673 (0.086) | ||||
| IL-12p70 | Higher in SRNS | NR; NR; 0.662 (0.108) | ||||
| IL-8 | Higher in SRNS | NR; NR; 0.638 (0.171) | ||||
| VEGF | Higher in SRNS | NR; NR; 0.638 (0.171) | ||||
| IL-13 | Higher in SRNS | NR; NR; 0.606 (0.294) | ||||
| IL-5 | Higher in SRNS | NR; NR; 0.569 (0.494) | ||||
| G-CSF | Higher in SRNS | NR; NR; 0.561 (0.546) | ||||
| MCP-1 | Lower in SRNS | NR; NR; 0.551 (0.611) | ||||
| FGF basic | Lower in SRNS | NR; NR; 0.545 (0.656) | ||||
| IP-10 | Lower in SRNS | NR; NR; 0.538 (0.703) | ||||
| IL-10 | Lower in SRNS | NR; NR; 0.524 (0.811) | ||||
| Eotaxin | Lower in SRNS | NR; NR; 0.452 (0.633) | ||||
| IL-4 | Lower in SRNS | NR; NR; 0.426 (0.464) | ||||
| PDGF-BB | Lower in SRNS | NR; NR; 0.385 (0.252) | ||||
| IL-17A | Lower in SRNS | NR; NR; 0.324 (0.08) | ||||
| RANTES | Lower in SRNS | NR; NR; 0.314 (0.065) | ||||
| IL-1ra | Lower in SRNS | NR; NR; 0.309 (0.058) | ||||
| MIP-1β | Lower in SRNS | NR; NR; 0.269 (0.022) | ||||
| IL-2 | Lower in SRNS | NR; NR; 0.260 (0.017) | ||||
| IL-9 | Lower in SRNS | NR; NR; 0.250 (0.013) | ||||
| Cuzzoni et al, 2019 (Italy)15 | Cohort, 62 (9/24+29c) | Pre-treatment, 6 weeks | Blood, Magnetic Beads Suspension Array (Bio-Plex Pro Human Cytokine 21- and 27-plex panels) | SCGF-β | Higher in SRNS (OR: 5.5; 95% CI: 1.1–56.4) | NR; NR; NR (0.034) |
| MIF | Higher in SRNS (OR: 4.3; 95% CI: 1.2–25.4) | 85.7; 71.4; 0.760 (0.022) | ||||
| Gooding et al, 2020 (USA)23 | Cohort, 41a (14/27) | Pre- and post-treatmentb, 7 weeks | Blood, Proton Nuclear Magnetic Resonance spectroscopy | Creatinine | Higher in SRNS (OR: 1.07; 95% CI: 1.02–1.12) | NR; NR; 0.79 (NR) |
| Glutamine | Higher in SRNS (OR: 1.01; 95% CI: 1.00–1.02) | NR; NR; 0.81 (NR) | ||||
| Malonate | Lower in SRNS for children >3 years (OR: 0.94; 95% CI: 0.89–1.00) | NR; NR; 0.84 (NR) | ||||
| Gopal et al, 2016 (India)20 | Cohort, 70 (23/47) | Pre-treatment, 18 weeks | Urine, Aminoff’s Method | uPBSA | Higher in SRNS | 75; 75.5; 0.814 (0.009) |
| Li et al, 2022 (China)16 | Cohort, 129 (35/94) | Pre-treatment, 4 weeks | Urine, NanoString nCounter and qRT-PCR | CREBBP | Higher in SRNS | 83.3; 49.4; 0.665 (0.007) |
| CYBA | Lower in SRNS | 58.3; 83.1; 0.669 (0.003) | ||||
| CREBBP + CYBA | Combination of CREBBP and CYBA is more effective to predict steroid resistance. | 63.9; 76.4; 0.666 (NR) | ||||
| Nickavar et al, 2016 (Iran)17 |
Cohort, 52 (27/25) | Pre-treatment, 8 weeks | Urine, ELISA | uNGAL/Creatinine Ratio | The optimal cut-off value for urine NGAL/Cr with the highest sensitivity and specificity was 0.46 ng/mg. | 63; 92; 0.759 (NR) |
| Pradeepchandran et al, 2022 (India)19 | Cohort, 79 (7/72) | Pre-treatment, 8 weeks | Urine, ELISA | uNGAL | uNGAL >10 ng/mL is associated with SRNS. | 100d; 87.5d; NR (<0.001) |
| Salah et al, 2023 (Egypt)25 | Cohort, 60 (19/41) | Pre-treatment, 4–8 weeks | Urine, ELISA | Urinary VDB | No significant difference between SSNS and SRNS at baseline. Levels reflect severity, not responsiveness. | NR; NR; NR (0.417) |
| Widiasta et al, 2025 (Indonesia)22 | Cohort, 24 (12/12) | Pre-treatment, 4 weeks | Blood, qT-PCR SYBR Green | TGF-β2 mRNA | Higher in SRNS | NR; NR; 0.972 (<0.001) |
| Blood, qT-PCR TaqMan miRNA Assays | miR-433 | Higher in SRNS | NR; NR; NR (0.03) | |||
| Zaorska et al, 2021 (Poland)18 | Cohort, 124 (53/71) | Pre-treatment, 8 weeks | Blood, Neural Network (Multivariate Model) | Prediction Model Panel of 9 SNPs + 2 Mutations + 1 CNV + 2 Methylation Status + 3 Demographics | ||
| ABCB1 rs1922240 + ABCB1 rs1045642 + ABCB1 rs2235048 + MIF rs2070767 + MIF rs2000466 + MIF rs5844572 + GLCCI1 rs37972 + NOTCH1 rs3124591 + CD73 rs9444348 + CD73 rs4431401 + EMP2 rs587777481 + NPHS2 rs1057516414 + SOCS3.1 + SOCS3.2 + Sex + Age of Onset + Serum Creatinine | Steroid resistance prediction reached 92% in binomial and 63% in multinomial calculations. | 33; 100; 0.778 (NR) | ||||
| Blood, Methylation-specific PCR | Methylation Status of SOCS3 Promoter Fragments | |||||
| SOCS3.1 promoter partial methylation | Less frequent in SRNS (OR: 0.7; 95% CI: 0.1–3.7). | NR; NR; NR (0.635) | ||||
| SOCS3.2 promoter unmethylation | 15-fold more frequent in SRNS (OR: 15.1; 95% CI: 5.8–39.3). | NR; NR; NR (<0.0001) | ||||
| Blood, PCR-RFLP/SNaPshot genotyping | Haplotype Frequencies | |||||
| ABCB1 (rs1922240 + rs1045642 + rs2235048) haplotype: AAC | More frequent in SRNS (OR: 1.4; 95% CI: 0.8–2.4). | NR; NR; NR (0.21) | ||||
| ABCB1 (rs1922240 + rs1045642 + rs2235048) haplotype: AGT | More frequent in SRNS (OR: 1.4; 95% CI: 0.7–2.7). | NR; NR; NR (0.282) | ||||
| ABCB1 (rs1922240 + rs1045642 + rs2235048) haplotype: GAC | More frequent in SRNS (OR: 1.2; 95% CI: 0.6–2.4). | NR; NR; NR (0.500) | ||||
| ABCB1 (rs1922240 + rs1045642 + rs2235048) haplotype: GGT | Less frequent in SRNS (OR: 0.5; 95% CI: 0.3–0.8). | NR; NR; NR (0.011) | ||||
| ABCB1 (rs1922240 + rs1045642 + rs2235048) haplotype: AGC | Less frequent in SRNS (OR: 0.001; 95% CI: 0–0.2). | NR; NR; NR (0.219) | ||||
| ABCB1 (rs1922240 + rs1045642 + rs2235048) haplotype: GGC | More frequent in SRNS (OR: 148.5; 95% CI: 6.1–3605). | NR; NR; NR (0.360) | ||||
| CD73 (rs9444348 + rs4431401) haplotype: AG | More frequent in SRNS (OR: 1.7; 95% CI: 1–2.8). | NR; NR; NR (0.046) | ||||
| CD73 (rs9444348 + rs4431401) haplotype: GA | Less frequent in SRNS (OR: 0.7; 95% CI: 0.4–1.2). | NR; NR; NR (0.180) | ||||
| CD73 (rs9444348 + rs4431401) haplotype: GG | Less frequent in SRNS. | NR; NR; NR (0.032) | ||||
| CD73 (rs9444348 + rs4431401) haplotype: AA | More frequent in SRNS. | NR; NR; NR (0.246) | ||||
| MIF (rs5844572 + rs2070767 + rs2000466) haplotype: 5AT | Less frequent in SRNS (OR: 1; 95% CI: 0.6–1.8). | NR; NR; NR (0.990) | ||||
| MIF (rs5844572 + rs2070767 + rs2000466) haplotype: 5GT | Less frequent in SRNS (OR: 0.3; 95% CI: 0.04–3). | NR; NR; NR (0.310) | ||||
| MIF (rs5844572 + rs2070767 + rs2000466) haplotype: 6GG | Less frequent in SRNS (OR: 0.8; 95% CI: 0.3–2.7). | NR; NR; NR (0.772) | ||||
| MIF (rs5844572 + rs2070767 + rs2000466) haplotype: 6GT | Less frequent in SRNS (OR: 1; 95% CI: 0.6–1.7). | NR; NR; NR (0.964) | ||||
| MIF (rs5844572 + rs2070767 + rs2000466) haplotype: 7GG | More frequent in SRNS (OR: 1.1; 95% CI: 0.5–2.2). | NR; NR; NR (0.883) | ||||
| MIF (rs5844572 + rs2070767 + rs2000466) haplotype: 6AT | More frequent in SRNS. | NR; NR; NR (0.098) | ||||
| MIF (rs5844572 + rs2070767 + rs2000466) haplotype: 7GT | More frequent in SRNS. | NR; NR; NR (0.237) | ||||
Notes: aOverlapping cohort for both studies; bOnly pre-treatment data is included for prediction analysis; cSteroid-dependent population; dIs counted manually by the reviewer, not reported by the author.
Abbreviations: AUC, area under the curve; NR, not reported; VDB, vitamin-D binding protein; ADIPOQ, adiponectin; MMP2, 72 kDA type IV collagenase; COL6A3, collagen alpha-3(VI) chain; IGFBP2, insulin-like growth factor binding protein 2; APOE, apolipoprotein E; SHBG, sex hormone-binding globulin; EFEMP1, epidermal growth factor-containing fibulin-like extracellular matrix protein 1; ITIH4, inter-alpha-trypsin inhibitor heavy chain H4; HPX, hemopexin; SERPINC1, antithrombin-III; AZGP1, zinc-alpha-2-glycoprotein; FETUB, fetuin-B; IL, interleukin; MCP, monocyte chemoattractant protein; IFN, interferon; MIP, macrophage inflammatory protein; TNF, tumor necrosis factor; VEGF, vascular endothelial growth factor; G-CSF, granulocyte colony-stimulating factor; FGF, fibroblast growth factor; IP, interferon inducible protein; PDGF-BB, platelet-derived growth factor B subunit homodimer; RANTES, regulated upon activation normal T cell expressed and presumably secreted; SCGF, stem cell growth factor; MIF, macrophage migration inhibitory factor; uPBSA, urinary protein-based sialic acid; CREBBP, cAMP response element-binding binding protein; CYBA, cytochrome b-245 alpha chain; uNGAL, urinary neutrophil gelatinase-associated lipocalin; miR, microRNA; SNP, single nucleotide polymorphism; CNV, copy number variation; ABCB, ATP-binding cassette subfamily B member 1; GLCCI1, glucocorticoid-induced transcript 1; NOTCH1, neurogenic locus notch homolog protein 1; CD73, cluster of differentiation 73; EMP2, epithelial membrane protein 2; NPHS2, nephrosis 2 idiopathic steroid-resistant (podocin); SOCS, suppressor of cytokine signaling.
The eleven included studies comprised 718 study-level observations. However, two studies from a single United States research group were derived from an overlapping parent cohort.23,24
The included studies spanned multiple levels of molecular biomarkers, encompassing genetic/genomic (eg, single nucleotide polymorphisms in ABCB1, CD73, and MIF), epigenetic (eg, SOCS3 promoter methylation), transcriptomic (eg, TGF-β mRNA, miR-433), proteomic (eg, plasma protein panels, cytokines), and metabolomic markers (eg, creatinine, glutamine, malonate). This categorization reflects the biological level at which each candidate biomarker originates and is used consistently throughout this review to distinguish predictive, diagnostic, and prognostic utility across marker classes.
Quality Assessment
Methodological quality of the included studies was assessed using the Newcastle-Ottawa Scale (NOS) for cohort studies, which evaluates three domains: selection of study groups (maximum 4 stars), comparability between groups (maximum 2 stars), and ascertainment of outcome (maximum 3 stars), yielding a maximum total score of 9 stars (Supplementary Table 2). Applying AHRQ standards, three studies achieved good quality (selection 3–4 stars, comparability 1–2 stars, and outcome 2–3 stars), whereas the remaining eight studies were classified as poor quality, as they scored 0 stars in the comparability domain despite otherwise adequate selection and outcome domains. Total scores ranged from 7 to 8 out of 9.
Overall, included studies demonstrated consistent strengths in cohort selection and outcome ascertainment, with all studies clearly defining steroid resistance based on standardized clinical response criteria (typically persistent proteinuria after 4–8 weeks of glucocorticoid therapy) and demonstrating that the outcome of interest was absent at baseline. However, a notable methodological limitation emerged in the comparability domain: 8 of 11 studies (72.73%) did not adjust for potential confounders, most notably age at disease onset (a well-established determinant of steroid responsiveness) when evaluating the association between the candidate biomarker and steroid resistance outcome.16,17,19–22,24,25 In these studies, biomarker levels were typically compared between SSNS and SRNS groups using unadjusted univariate tests (eg, Student’s t-test, Mann–Whitney U), despite baseline demographic differences (particularly age) being present or unreported. This resulted in lack of comparability-domain stars and a total NOS score of 7/9 for each of these studies.
One additional study also scored 7/9.15 However, its methodological limitation was distinct, reflecting inadequate follow-up duration rather than a comparability deficiency, as this study appropriately controlled for age as a confounder in its multivariate logistic regression model when identifying MIF as a predictor of steroid resistance. The two highest-scoring studies by contrast, achieved the full 8/9 by incorporating age-adjusted or multivariate modeling approaches when assessing their respective biomarker panels against steroid resistance outcomes.18,23
These findings suggest that while the existing SRNS biomarker literature generally reflects adequate cohort design and outcome definition, the limited control for confounding (particularly age), represents a systematic methodological gap that should be considered when interpreting the independent predictive value of the biomarkers identified across these studies.
Molecular Biomarkers for Predicting SRNS
Eleven cohort studies evaluated a diverse range of molecular biomarkers; spanning proteomic, cytokine, metabolomic, urinary protein, transcriptomic, and genetic/epigenetic domains for their ability to predict or differentiate SRNS from SSNS in pediatric patients (Table 1).
Proteomic Biomarkers
Agrawal et al (2020) identified 13 plasma proteins that individually differed significantly between SRNS and SSNS at disease presentation, including vitamin-D binding protein (VDB), hemopexin (HPX), and adiponectin (ADIPOQ) with all p-values <0.05.24 A three-protein panel combining VDB, ADIPOQ, and matrix metalloproteinase 2 (MMP2) achieved an AUC of 0.78 (p = 0.003) for predicting SRNS at onset, outperforming any single biomarker alone.
Cytokine Biomarkers
Agrawal et al (2021) screened 27 plasma cytokines and found interferon-γ (IFN-γ) to have the strongest individual discriminative power (AUC 0.792, p = 0.004), followed by MIP-1α, IL-6, and IL-1β (all AUC > 0.72, p < 0.03).21 A composite panel of IL-7, IL-9, and MCP-1 improved prediction substantially, yielding an AUC of 0.846 (p < 0.001). In a separate cohort, Cuzzoni et al (2019) reported that macrophage migration inhibitory factor (MIF) was significantly elevated in SRNS patients (OR 4.3, 95% CI 1.2–25.4, p = 0.022), with a cutoff of 501 pg/mL yielding 85.7% sensitivity and 71.4% specificity (AUC 0.760); SCGF-β was also elevated (OR 5.5, 95% CI 1.1–56.4, p = 0.034).15
Metabolomic Biomarkers
Gooding et al (2020) used plasma metabolomic profiling. In an analysis of children of all ages (n = 39), creatinine alone predicted SRNS (OR 1.07, 95% CI 1.02–1.12, AUC 0.79); when age was entered into the model, glutamine replaced creatinine as the selected metabolite, and the combined age-plus-glutamine model reached an AUC of 0.810 (glutamine OR 1.01, 95% CI 1.00–1.02), compared with 0.77 for age alone. In a separate analysis restricted to children older than 3 years (n = 28), a combined age-plus-malonate model reached an AUC of 0.84 (malonate OR 0.94, 95% CI 0.89–1.00), compared with 0.72 for age alone.23
Urinary Protein Biomarkers
Several studies evaluated urinary markers reflecting oxidative stress, tubular injury, or renal filtration barrier integrity. Gopal et al (2016) found urinary protein-bound sialic acid (uPBSA) to differentiate SRNS from SSNS with 75% sensitivity, 75.5% specificity, and an AUC of 0.814 (p = 0.009).20 Two independent studies evaluated urinary neutrophil gelatinase-associated lipocalin (uNGAL): Nickavar et al (2016) established an optimal uNGAL/creatinine cutoff of 0.46 ng/mg (63% sensitivity, 92% specificity, AUC 0.759), while Pradeepchandran et al (2022) found an uNGAL threshold of >10 ng/mL strongly associated with SRNS (100% sensitivity, 87.5% specificity, p < 0.001).17,19 In contrast, Salah et al (2023) found no significant difference in urinary vitamin D-binding protein levels between SSNS and SRNS at baseline (p = 0.417), concluding that this marker reflects disease severity rather than steroid responsiveness.25
Transcriptomic and Epigenetic Biomarkers
Li et al (2022) evaluated urinary sediment mRNA levels of CREBBP and CYBA, reporting individual AUCs of 0.665 (p = 0.007) and 0.669 (p = 0.003), respectively; the combined panel yielded a comparable AUC of 0.666.16 Widiasta et al (2025) demonstrated that blood TGF-β2 mRNA expression was markedly elevated in SRNS (AUC 0.972, p <0.001), while miR-433 was also significantly upregulated (p = 0.03), with no other assessed miRNAs (miR-21, miR-29, miR-200, miR-205) reaching statistical significance.22
Genetic and Epigenetic Prediction Models
Zaorska et al (2021) developed the most complex prediction model, integrating nine single nucleotide polymorphisms, two mutations, one copy number variant, methylation status of the SOCS3 promoter, and three demographic variables (sex, age of onset, serum creatinine).18 This composite model achieved 92% accuracy in binomial classification and an AUC of 0.778 (33% sensitivity, 100% specificity). Among individual epigenetic findings, unmethylation of the SOCS3.2 promoter fragment was markedly more frequent in SRNS (OR 15.1, 95% CI 5.8–39.3, p <0.0001), representing the strongest single association identified across all included studies.
Collectively, these findings indicate that while numerous individual biomarkers demonstrate statistically significant associations with SRNS, composite panels generally outperformed most individual markers (panel AUC range 0.666–0.846), with the notable exception of TGF-β2 mRNA (AUC 0.972), the single highest-performing biomarker identified overall, albeit from a very small cohort (n = 24) discussed further below regarding overfitting risk.
Discussion
This systematic review synthesizes current evidence on non-invasive molecular biomarkers for predicting steroid resistance in pediatric nephrotic syndrome. It is worth noting that this review’s scope, limited to non-invasive molecular biomarkers in pre-treatment cohort designs, does not capture parallel advances from large pediatric nephrology consortia such as PodoNet and NEPTUNE, nor the expanding role of genetic panels in identifying monogenic causes of SRNS, both of which represent complementary avenues toward precision diagnosis. As emphasized in the introduction, the wait-and-see approach often exposes children to unnecessary steroid toxicity.8,9 Several included studies reported that combining multiple cytokines or plasma proteins into a single panel improved discriminative accuracy over any individual analyte: Agrawal et al (2021) reported an AUC of 0.846 for a three-cytokine panel (IL-7, IL-9, MCP-1), and Agrawal et al (2020) reported an AUC of 0.780 for a three-protein panel (VDB, ADIPOQ, MMP2).21,24 A related single-cytokine finding by Cuzzoni et al (2019) identified macrophage migration inhibitory factor (MIF) as individually predictive of steroid resistance (AUC 0.760) in a discovery cohort of 21 patients.15 Notably, a parallel cutoff identified for steroid-dependent patients in the same discovery cohort could not be confirmed when tested in the larger extended cohort, illustrating the risk of unstable cutoffs derived from small discovery samples, a caution that applies broadly to several biomarkers discussed in this review. Taken together, these findings are hypothesis-generating for future risk-stratification approaches, though translation into early clinical decision-making would require prospective validation beyond the exploratory cohorts reported here.
The biomarkers identified also offer plausible, though unconfirmed, links to proposed pathophysiological differences between SRNS and SSNS. The upregulation of TGF-β2 mRNA and miR-433 reported by Widiasta et al (2025) is consistent with the hypothesis that SRNS may involve early activation of pro-fibrotic, podocyte-associated pathways rather than solely transient immune dysregulation; however, this remains a single small cohort finding rather than an established mechanism.22,26 Similarly, the very high discriminative value reported for TGFβ2 (AUC 0.972, n = 24) should be interpreted cautiously, as such an estimate in a small, single-center cohort is susceptible to overfitting and requires independent replication before any conclusion about early glomerular remodelling can be drawn.22,27,28 A further, non-mutually-exclusive interpretation is that several of the identified biomarkers may reflect an underlying histopathological transition from minimal change disease to focal segmental glomerulosclerosis (FSGS), a transition long recognized since the original International Study of Kidney Disease in Children (ISKDC) cohort, to correlate with steroid resistance, rather than steroid responsiveness per se.3,5,29 This framework is broadly consistent with the pro-fibrotic signal captured by TGF-β2 mRNA and miR-433 and with proteomic markers of extracellular matrix remodelling such as MMP2.22,24 However, none of the included studies performed a paired or sequential histopathological assessment, so this possibility remains speculative and was not directly tested within this review. The elevation of creatinine and glutamine reported by Gooding et al (2020) may reflect early metabolic differences between SRNS and SSNS patients even when eGFR appears preserved, though whether this reflects an underlying mechanism such as mitochondrial dysfunction or altered renal functional reserve remains speculative and was not directly tested in the included studies.23,30 We frame these interpretations explicitly as hypotheses warranting mechanistic investigations, rather than as demonstrated disease features.
An important observation from this review concerns the heterogeneous performance of urinary biomarkers as a class. While plasma VDBP contributed to a multi-protein predictive panel in one cohort, its urinary counterpart showed inconsistent performance: Salah et al found no predictive value for urinary VDBP in their cohort, whereas earlier studies by Bennett et al, and Choudhary et al reported that urinary VDBP could distinguish SRNS from SSNS.2,25,31 This inconsistency may reflect differences in study design (cross-sectional in Bennett et al and Choudhary et al versus longitudinal in Salah et al), confounding of urinary biomarkers by non-selective proteinuria during active disease, or population-specific heterogeneity.2,25,31 This inconsistency, however, does not extend uniformly to urinary biomarkers as a whole. Several other urinary candidates showed statistically significant and, in some cases, high-accuracy discrimination between SRNS and SSNS: urinary protein-bound sialic acid (Gopal et al 2016; AUC 0.814) reflecting oxidative stress pathways, and urinary NGAL, evaluated independently by Nickavar et al (2016; AUC 0.759) and Pradeepchandran et al (2022; 100% sensitivity, 87.5% specificity), reflecting tubular injury.17,19,20 Li et al (2022) additionally reported that urinary sediment mRNA levels of CREBBP and CYBA discriminated SRNS with modest individual accuracy (AUC 0.665–0.669).16 Rather than concluding that systemic biomarkers are inherently superior to urinary biomarkers, we note that biomarker performance appears to vary more by the specific analyte and its underlying biological pathway (oxidative stress, tubular injury, or glomerular permeability) than by specimen type alone; direct head-to-head comparisons within the same cohort would be needed to draw firmer conclusions about the relative value of plasma versus urinary approaches.
A distinct modeling approach was reported by Zaorska et al (2021), who integrated nine single nucleotide polymorphisms, one copy number variant, SOCS3 promoter methylation status, and clinical/demographic variables into a composite neural network model, achieving 92% accuracy in binomial classification (AUC 0.778).18 Within this model, unmethylation of the SOCS3.2 promoter fragment showed the strongest individual association with SRNS (OR 15.1, 95% CI 5.8–39.3).18 This finding illustrates a genetic/epigenetic approach that is conceptually distinct from the plasma- and urine-based analyte panel discussed above, combining heritable variation with an environmentally responsive epigenetic mark.18 While promising, this model’s complexity, which requires genotyping, methylation assays, and multiple demographic inputs simultaneously, in our assessment may currently limit feasibility for routine clinical translation compared with single-analyte or smaller composite panels, an important practical consideration alongside its promising discriminative performance.
These findings support a potential future adjunctive role for molecular biomarkers in the risk stratification of pediatric nephrotic syndrome, contingent on further validation. At this stage, however, the evidence does not support using a high-risk molecular profile (such as elevated TGF-β2, a specific cytokine panel, or a genetic/epigenetic composite score) to bypass the standard glucocorticoid trial in favor of upfront genetic testing or second-line immunosuppression; such an approach would be inconsistent with current IPNA and KDIGO recommendations, which continue to rely on clinical response to a standardized steroid trial for the SSNS/SRNS distinction.8,9 Future research should prioritize validating these candidate biomarkers in larger, prospective, multicenter cohorts with standardized assay platforms and pre-specified diagnostic thresholds, a necessary prerequisite before any consideration of a bedside molecular risk score.
Beyond molecular biomarkers, the KDIGO 2025 guideline also recommends early genetic testing in children with SRNS, particularly those with a family history, extrarenal features, or early-onset disease, as identifying monogenic causes can both confirm diagnosis and obviate unnecessary immunosuppressive exposure.9 For children meeting these established criteria, genetic testing therefore remains the primary validated tool for early risk stratification in this population; the biomarkers discussed here, including the genetic/epigenetic model of Zaorska et al may eventually offer a complementary, less invasive adjunct, but do not currently substitute for clinically indicated genetic evaluation.
Although this systematic review identifies several molecular candidates of interest for future study in pediatric nephrotic syndrome, these findings must be interpreted in light of substantial limitations. First, although the search strategy identified 11 eligible cohort studies for inclusion, the evidence base remains limited in absolute terms, with individual cohort sample sizes ranging from 24 to 129 patients. Given these small sample sizes, the reported AUC values, particularly the exceptionally high estimate for TGF-β2 mRNA (AUC 0.972) reported in a cohort of only 24 patients, may be subject to small-sample optimism and model overfitting, and should be interpreted as preliminary until replicated in independent, adequately powered cohorts.22 Notably, three of the eleven included studies (Agrawal et al 2020, Agrawal et al 2021, and Gooding et al 2020) originated from the same research consortium (the Midwest Pediatric Nephrology Consortium, subsequently renamed the Pediatric Nephrology Research Consortium), which may introduce cohort- or population-specific bias and limit independent replication of these particular findings. Critically, the cohorts reported by Agrawal et al (2020) and Gooding et al (2020) were explicitly confirmed by the original authors to derive from the same parent patient population; Agrawal et al (2021) was conducted through the same consortium infrastructure, though patient-level overlap with the other cohorts cannot be confirmed from the published reports. Because of this overlap, the total of 718 participants summed across all 11 included studies should be interpreted as a sum of study-level observations rather than a count of unique individual children; the true number of unique children contributing to this evidence base is very likely lower, though it cannot be precisely determined from the available published data.
Second, there is substantial geographical and ethnic heterogeneity among the included studies, which comprised cohorts from the United States, Italy, India, Iran, China, Egypt, Indonesia, and Poland. Given that genetic background and ethnicity are known to influence the epidemiology and pathophysiology of nephrotic syndrome, this heterogeneity could introduce genetic confounders affecting biomarker expression profiles and limit the direct comparability of effect estimates across cohorts.
Third, substantial methodological heterogeneity was observed in the detection techniques employed, ranging from enzyme-linked immunosorbent assays (ELISA) and multiplex bead-based immunoassays to proteomic and metabolomic profiling (LC-MS, NMR), qRT-PCR, and genotyping/methylation-based approaches. This variation affects assay sensitivity and specificity, combined with the diverse biological matrices used (blood and urine) and heterogenous outcome-ascertainment windows (4 to 18 weeks), precludes a quantitative meta-analysis at this stage. Accordingly, while several candidate biomarkers identified in this review are promising, all require validation in larger, multicenter cohorts using standardized methodologies before any consideration of clinical implementation.
An additional limitation concerns the comprehensiveness of the search strategy. The databases accessible to the authors were limited to PubMed, Scopus, ScienceDirect, and MEDLINE Ultimate via EBSCO; the authors did not have institutional access to Web of Science, EMBASE, or CINAHL. While backward citation searching was performed to mitigate this constraint, additional eligible studies indexed exclusively in these databases may not have been captured. This limitation reflects institutional access constraints rather than a deliberate restriction in scope.
Finally, we note that the included studies vary in the specific clinical role tested for each biomarker: most, including Agrawal et al (2020, 2021), Gooding et al (2020), and Cuzzoni et al (2019), evaluated biomarkers in a predictive capacity, measured before treatment to forecast subsequent steroid response, whereas others, such as the genetic/epigenetic model of Zaorska et al combined predictive and mechanistic elements within a single composite score. The conclusions of this review are accordingly limited to the biomarkers’ demonstrated predictive associations with subsequent steroid response in small, exploratory cohorts, rather than established diagnostic or prognostic utility, pending independent validation.
Conclusion
Current evidence from eleven pediatric cohort studies suggests that selected molecular biomarkers; including cytokine and protein panels, urinary markers of oxidative stress and tubular injury, transcriptomic markers such as TGF-β2 mRNA expression, and a composite genetic/epigenetic model incorporating single nucleotide polymorphisms and SOCS3 promoter methylation; show statistically significant associations with subsequent steroid resistance in children with nephrotic syndrome. These findings are hypothesis-generating and should be interpreted as exploratory, given the small size and heterogeneity of individual cohorts, the concentration of two of the eleven included studies within one research group with likely overlapping participants, and substantial variability in detection platforms and biological matrices across studies. The available evidence does not establish that SRNS is definitively molecularly distinct from SSNS at presentation, nor that any single biomarker class demonstrates superior predictive value over others; such comparative claims would require direct head-to-head evaluation within the same cohorts, which none of the included studies performed. An alternative interpretation worth considering is that the candidate biomarkers identified here may not primarily reflect steroid responsiveness per se, but rather an underlying tendency of nephrotic syndrome to transition histopathologically from minimal change disease toward focal segmental glomerulosclerosis; a transition long recognized to correlate with steroid resistance. This possibility remains speculative and was not directly tested by the included studies, none of which performed paired or sequential histopathological assessment, but it offers a plausible mechanistic framework for future research linking molecular signatures to underlying glomerular pathology rather than to treatment response in isolation. Rather than supporting immediate clinical translation, these results highlight candidate molecular signatures warranting validation in prospective, multicenter cohorts using pre-specified biomarker panels, standardized assay platform, and independent test sets. To date, no biomarker has been shown capable of reliably screening, at disease onset, whether a given case of nephrotic syndrome will prove steroid-resistant or steroid-sensitive; at present, these biomarkers have not been validated for use in place of standard steroid trials, nor to guide upfront genetic testing or immunosuppressive treatment decisions, and clinical decision-making should continue to follow current IPNA and KDIGO recommendations pending further evidence.
Acknowledgments
This publication charge is funded by Unpad through the Indonesian Endowment Fund for Education (LPDP) on behalf of the Indonesian Ministry of Higher Education, Science and Technology and managed under the EQUITY Program (Contract No. 4303/ B3/DT.03.08/2025 and 3927/UN6. RKT/HK.07.00/2025).
Generative AI from Gemini Thinking with 3 Pro Model is used in order to refine the grammar, language, and readability of this manuscript.
Disclosure
Ahmedz Widiasta is an author of one of the studies included in this review.22 To manage this conflict, Ahmedz Widiasta was not involved in the data extraction or quality assessment of that specific study. The authors report no conflicts of interest in this work.
References
- 1.Chanchlani R, Parekh RS. Ethnic differences in childhood nephrotic syndrome. Front Pediatr. 2016;4(APR). doi: 10.3389/fped.2016.00039 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Stone H, Magella B, Bennett MR. The search for biomarkers to aid in diagnosis, differentiation, and prognosis of childhood idiopathic nephrotic syndrome. Front Pediatr. 2019;7. doi: 10.3389/fped.2019.00404. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Nourbakhsh N, Mak R. Steroid-resistant nephrotic syndrome: past and current perspectives. Pediatric Health Med Ther. 2017;8:29–15. doi: 10.2147/phmt.s100803 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Saleem MA. Molecular stratification of idiopathic nephrotic syndrome. Nat Rev Nephrol. 2019;15(12):750–765. doi: 10.1038/s41581-019-0217-5 [DOI] [PubMed] [Google Scholar]
- 5.Becherucci F, Landini S, Cirillo L, Mazzinghi B, Romagnani P. Look alike, sound alike: phenocopies in steroid-resistant nephrotic syndrome. Int J Environ Res Public. 2020;17(22):1–17. doi: 10.3390/ijerph17228363 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Becherucci F, Roperto RM, Materassi M, Romagnani P. Chronic kidney disease in children. Clin Kidney J. 2016;9(4):583–591. doi: 10.1093/ckj/sfw047 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Uwaezuoke SN. The role of novel biomarkers in childhood idiopathic nephrotic syndrome: a narrative review of published evidence. Int J Nephrol Renovasc Dis. 2017;10:123–128. doi: 10.2147/IJNRD.S131869 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Trautmann A, Vivarelli M, Samuel S, et al. IPNA clinical practice recommendations for the diagnosis and management of children with steroid-resistant nephrotic syndrome. Pediatr Nephrol. 2020;35(8):1529–1561. doi: 10.1007/s00467-020-04519-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Floege J, Gibson KL, Vivarelli M, Liew A, Radhakrishnan J, Rovin BH. KDIGO 2025 clinical practice guideline for the management of nephrotic syndrome in children. Kidney Int. 2025;107(5):S241–S289. doi: 10.1016/j.kint.2024.11.007 [DOI] [PubMed] [Google Scholar]
- 10.Zhang H, Wang Z, Dong LQ, Yn GUO. Children with steroid-resistant nephrotic syndrome: long-term outcomes of sequential steroid therapy. Biomed Environ Sci. 2016;29(9):650–655. doi: 10.3967/bes2016.087 [DOI] [PubMed] [Google Scholar]
- 11.Trachtman H, Desmond H, Williams AL, et al. Rationale and design of the nephrotic syndrome study network (Neptune) Match in glomerular diseases: designing the right trial for the right patient, today. Kidney Int. 2024;105(2):218–230. doi: 10.1016/j.kint.2023.11.018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.May CJ, Ford NP, Welsh GI, Saleem MA. Biomarkers to predict or measure steroid resistance in idiopathic nephrotic syndrome: a systematic review. PLoS One. 2025;20(2 February). doi: 10.1371/journal.pone.0312232 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021:372. doi: 10.1136/bmj.n71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Haddaway NR, Page MJ, Pritchard CC, McGuinness LA. PRISMA2020: an R package and Shiny app for producing PRISMA 2020-compliant flow diagrams, with interactivity for optimised digital transparency and open synthesis. Campbell Syst Rev. 2022;18(2):e1230. doi: 10.1002/cl2.1230 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Cuzzoni E, Franca R, De Iudicibus S, et al. MIF plasma level as a possible tool to predict steroid responsiveness in children with idiopathic nephrotic syndrome. Eur J Clin Pharmacol. 2019;75(12):1675–1683. doi: 10.1007/s00228-019-02749-3 [DOI] [PubMed] [Google Scholar]
- 16.Li W, Shou X, Xiang W, et al. Urinary sediment mRNA level of CREBBP and CYBA in children with steroid-resistant nephrotic syndrome. Front Immunol. 2022:12. doi: 10.3389/fimmu.2021.801313. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Nickavar A, Safaeian B, Sadeghi-Bojd S, Lahouti Harah Dashti A. Urine neutrophil gelatinase associated lipocalin to creatinine ratio: a novel index for steroid response in idiopathic nephrotic syndrome. Ind J Pediatr. 2016;83(1):18–21. doi: 10.1007/s12098-015-1809-0 [DOI] [PubMed] [Google Scholar]
- 18.Zaorska K, Zawierucha P, Świerczewska M, Ostalska-Nowicka D, Zachwieja J, Nowicki M. Prediction of steroid resistance and steroid dependence in nephrotic syndrome children. J Transl Med. 2021;19(1). doi: 10.1186/s12967-021-02790-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Pradeepchandran G, Joseph S, Uthup S, Saradakutty G. Urine neutrophil gelatinase-associated lipocalin as a prognostic biomarker in the first episode of idiopathic nephrotic syndrome in children. J Pediatr Res. 2022;9(4):376–382. doi: 10.4274/jpr.galenos.2022.00008 [DOI] [Google Scholar]
- 20.Gopal N, Chandra Koner B, Bhattacharjee A, Bhat V. Assay of Urinary Protein-Bound Sialic Acid Can Differentiate Steroid-Sensitive Nephrotic Syndrome From Steroid-Resistant Cases. Vol. 27. 2016. [DOI] [PubMed] [Google Scholar]
- 21.Agrawal S, Brier ME, Kerlin BA, et al. Plasma cytokine profiling to predict steroid resistance in pediatric nephrotic syndrome. Kidney Int Rep. 2021;6(3):785–795. doi: 10.1016/j.ekir.2020.12.027 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Widiasta A, Sribudiani Y, Nugrahapraja H, Rachmadi D. miRNAs involved in the TGFB signaling as possible markers of steroid-resistant nephrotic syndrome in children. Gene Rep. 2025;39. doi: 10.1016/j.genrep.2025.102173. [DOI] [Google Scholar]
- 23.Gooding JR, Agrawal S, McRitchie S, et al. Predicting and defining steroid resistance in pediatric nephrotic syndrome using plasma metabolomics. Kidney Int Rep. 2020;5(1):81–93. doi: 10.1016/j.ekir.2019.09.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Agrawal S, Merchant ML, Kino J, et al. Predicting and defining steroid resistance in pediatric nephrotic syndrome using plasma proteomics. Kidney Int Rep. 2020;5(1):66–80. doi: 10.1016/j.ekir.2019.09.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Salah DM, Aoun AH, Fahmy YA, Fahmy BS, Zeid AHMA. Does urinary Vitamin D-binding protein have a role in the prediction of steroid resistance in nephrotic syndrome? A cohort study on Egyptian children. J Comprehens Pediatr. 2023;14(1). doi: 10.5812/compreped-130133 [DOI] [Google Scholar]
- 26.de Castro PAS V, Fujihara Ide T, Crespo Torres F, Simões e Silva AC. The view of pediatric nephrotic syndrome as a podocytopathy. Kidney Dial. 2023;3(4):346–372. doi: 10.3390/kidneydial3040030 [DOI] [Google Scholar]
- 27.Lim BJ, Yang JW, Do WS, Fogo AB. Pathogenesis of focal segmental glomerulosclerosis. J Pathol Transl Med. 2016;50(6):405–410. doi: 10.4132/jptm.2016.09.21 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Widiasta A, Wahyudi K, Sribudiani Y, Rachmadi D. The level of transforming growth factor-b as a possible predictor of cyclophosphamide response in children with steroid-resistant nephrotic syndrome. BioMedicine. 2021;11(3):68–75. doi: 10.37796/2211-8039.1205 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.International Study of Kidney Disease in Children. Nephrotic syndrome in children: prediction of histopathology from clinical and laboratory characteristics at time of diagnosis. Kidney Int. 1978;13(2):159–165. doi: 10.1038/ki.1978.23 [DOI] [PubMed] [Google Scholar]
- 30.Li Y, Fan J, Zhu W, Niu Y, Wu M, Zhang A. Therapeutic potential targeting podocyte mitochondrial dysfunction in focal segmental glomerulosclerosis. Kidney Dis S. 2023;9(4):254–264. doi: 10.1159/000530344 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Choudhary A, Mohanraj PS, Krishnamurthy S, Rajappa M. Association of Urinary Vitamin D Binding Protein and Neutrophil Gelatinase-Associated Lipocalin with Steroid Responsiveness in Idiopathic Nephrotic Syndrome of Childhood. Vol. 31. 2020. [DOI] [PubMed] [Google Scholar]