Kidney Survival in Children With Steroid-Resistant Nephrotic Syndrome Treated by Rituximab Authors: Eugene Yu-hin Chan, Irene Yuk-ying Ho, Chun-Ting Au, Ellen L.M. Yu, Andrea Angeletti, Sushmita Banerjee, Olivia Boyer, Chang-Yien Chan, Anna Francis, Gian Marco Ghiggeri, Riku Hamada, Nakysa Hooman, Luke Sydney Hopf, Mohamad Ikram I, Iftikhar Ijaz, Dmytro D. Ivanov, Hee Gyung Kang, Laura Lucchetti, Francesca Lugani, Alison Lap-tak Ma, William Morello, Subal Kumar Pradhan, Larisa Prikhodina, Reem H. Raafat, Rajiv Sinha, Sharon Teo, Kouki Tomari, Marina Vivarelli, Hazel Webb, Hui Kim Yap, Desmond Yat-hin Yap, Kjell Tullus Published: 2026 ABSTRACT IntroductionKidney outcomes in children with steroid-resistant nephrotic syndrome (SRNS) following rituximab remain unclear.MethodsWe conducted an international retrospective cohort study across 23 centers in 17 countries, including children with SRNS who did not respond to calcineurin inhibitors (CNIs) and subsequently, received rituximab. Patients with genetic variants were excluded. The primary outcome was kidney survival.ResultsWe analyzed data on 151 children (53% males; median age at onset: 6.8 years; primary vs. secondary SRNS: 54% vs. 46%; focal segmental glomerulosclerosis [FSGS]: 62%). All subjects received CNIs before rituximab (0–3 months: 21%; 3–6 months: 26%; 6–12 months: 26%; > 12 months: 26%). Upon rituximab, 73 subjects (48%) had normal kidney function, 47 (31%) had chronic kidney disease (CKD)2, and 31 (21%) had CKD3. Twenty-eight (19%) developed kidney failure. Overall kidney survival was 82.7%, 75.8%, and 72.3% at 3, 5, and 7 years post-rituximab. Baseline CKD staging before rituximab was associated with kidney survival at 5 years (log-rank P < 0.001); CKD stage 1, 91.7%; CKD stage 2, 70.4%; and CKD stage 3, 35.5%). The predictive factors for inferior kidney survival were CKD2 (adjusted hazard ratio [HRadj]: 2.7, 95% confidence interval [CI]: 1.5–4.9, P < 0.001), lower baseline serum albumin (log-rank P = 0.01; HRadj: 0.88, 95% CI: 0.81–0.95; P < 0.001) and FSGS (log-rank P = 0.03; HRadj: 3.4, 95% CI: 1.1–10.0; P < 0.001). Nonremission at 6 months was associated with inferior kidney survival (61.2% at 5 years), compared with complete remission (CR) or partial remission (PR) (100% at 5 years) (log-rank P = 0.01; HRadj: 14.1, 95% CI: 2.2–93.5, P = 0.01). Duration of prior calcineurin inhibition and SRNS type were not significant predictors.ConclusionKidney survival of SRNS following add-on rituximab is 70% to 80% over 3 to 7 years. Nonresponse, preexisting CKD, lower albumin, and FSGS predict inferior kidney survivals. SRNS is a rare disease in childhood, leading to considerable morbidity and mortality. Its pathogenesis primarily involves immune dysregulation and monogenic podocytopathies that disrupt the glomerular filtration membrane.1,2 In immune-mediated cases, intensified immunosuppression with CNIs is the recommended first-line therapy.3 Rituximab, though controversial, is frequently employed by pediatric nephrologists for nonresponding patients and may achieve remission in a subset of cases, with reported rates ranging from 30% to 80%.4,5 Although current guidelines define CNI resistance as failure after ≥6 months of therapy,3 early rituximab administration before this timeline is not uncommon in clinical practice, driven by indications, including disease-related complications, prolonged and repeated hospitalizations, intolerance to CNIs, physician preference, and parental anxiety.6 Children with SRNS, in contrast to steroid-sensitive NS, are at high risk of developing CKD and kidney failure.1 Previous reports consistently identified nonresponse to immunosuppression and genetic podocytopathies as key risk factors for kidney survival.7, 8, 9, 10 In contrast, the prognostic values of histological subtype and type of SRNS (primary vs. secondary) remain conflicting.7, 8, 9, 10, 11 The PodoNet Consortium reported long-term kidney survival in the largest pediatric SRNS cohort to-date, comprising 1354 children.8 However, this cohort constituted both immune-mediated and monogenic forms of SRNS. Notably, only 7 patients received rituximab therapy,8 which is considered a potential rescue therapy according to the current International Pediatric Nephrology Association guideline.3 Similar limitations were evident in other pediatric cohorts, where majority of children were treated with cyclosporine A and/or cyclophosphamide.7,9,12,13 In addition to these constraints, previous studies often included mixed adult populations or focused on specific histological subtypes, such as FSGS.14,15 Consequently, data on kidney survival among the most refractory immune-mediated SRNS cases is extremely scarce. We previously demonstrated that rituximab could improve remission status in an international cohort of children with SRNS unresponsive to CNIs, with remission correlating with kidney outcomes.6 To address the current knowledge gap, we conducted a secondary analysis to examine kidney survival in children with SRNS treated with rituximab, and to identify predictive factors that may inform prognosis. Methods We conducted an international retrospective observational cohort study, between June 1, 2021 and June 30, 2022, on children and adolescents with SRNS treated with rituximab. Twenty-three tertiary pediatric nephrology centers from 17 countries in Asia, Europe, North America, and Oceania participated in this secondary analysis. We used standard definitions and remission criteria for NS (Supplementary Table S1).1,23 We included pediatric patients with SRNS (aged 1–18 years at disease onset) treated with rituximab before the age of 21 years to induce remission. All patients had SRNS, which failed to achieve CR or PR after 4 weeks of prednisolone or prednisone at standard dose (60 mg/m2/d or 2 mg/kg/d). Primary steroid resistance (SR) occurred at first presentation, whereas secondary SR referred to children with initial steroid-responsiveness who became steroid-resistant during a subsequent relapse.1 All patients underwent kidney biopsy and received CNIs (cyclosporine A, 3–5 mg/kg/d; or tacrolimus, 0.05–0.2 mg/kg/d) for ≥1 month before rituximab. Rituximab (375–1500 mg/m2/course: 1–4 weekly infusions of 375 mg/m2 or 2 biweekly infusions of 750 mg/m2) was administered according to local policies between January 1, 2005 and December 31, 2021. Patients who received extracorporeal therapies (therapeutic plasma exchange, lipid apheresis, or immunoadsorption) or other biologics were excluded. Other exclusion criteria included congenital or infantile NS, genetic podocytopathies (pathological variants or Alport syndrome), secondary NS (e.g., diffuse mesangial sclerosis, C3 glomerulopathy, IgA nephropathy, etc.), posttransplant recurrence, and patients with inadequate data for analysis. Genetic tests were performed according to institutional practice and test availability, including Sanger, next generation, and/or whole exome sequencing. We collected data pertaining to patient demographics, kidney histology, rituximab and other treatments details, and kidney outcomes. All data collected by participating site investigators were fully anonymized and recorded on standardized forms. The primary outcome was survival free from kidney failure. Kidney failure was defined as initiation of maintenance kidney replacement therapy. Children were censored at last follow-up at the participating centers. Secondary outcomes included predictors for kidney survival, proteinuria, and estimated glomerular filtration rate (eGFR). Proteinuria was quantified by urine protein-to-creatinine ratio (UPCR) on the first-void morning urine sample. eGFR was calculated according to the revised bedside Schwartz formula and the CKD-Epidemiology Collaboration 2021 race-free formula for patients aged < 18 and ≥ 18 years, respectively.25,26 Normal kidney function was defined as eGFR ≥ 90 ml/min per 1.73 m2. Kidney histology was evaluated by pathologists from each participating institution. The study obtained ethical approval from the Research Ethics Committee of Hong Kong Children’s Hospital (HKCH-REC-2021-016), as the study lead site. Participating centers obtained relevant ethical approval in compliance with local regulations, where required, because some centers did not mandate such review for retrospective studies. Patient consent was waived because of the retrospective study nature. The study was conducted in compliance with the Declaration of Helsinki and the Declaration of Istanbul. Statistical Analyses Variables were compared between groups with and without kidney failure and/or mortality using independent t test, Mann-Whitney U test, Pearson’s chi-square test and Fisher exact test, as appropriate. Kidney survival data from the date of rituximab administration were analyzed using the Kaplan–Meier method. Survival curves were generated for all participants, as well as clinical and treatment subgroups, with follow-up truncated at 10 years to obtain comparable survival estimates. Group differences were assessed using the log-rank test, and survival estimates with 95% CIs were reported at clinically relevant time points. For remission status at 6 months, we performed a landmark analysis, defining time 0 set as 6 months from rituximab administration. Receiver operating characteristic analyses were performed to determine the optimal cutoffs for UPCR and serum albumin at the time of rituximab using Youden index to discriminate kidney failure from all patients. Because remission status following rituximab was a time-varying covariate, we used time-dependent Cox model to avoid immortal time bias. We determined predictors of kidney failure and/or mortality using Cox proportional hazards models with a robust variance estimator to take within-center correlations into account. Initially, we performed univariate analysis for potential variables. These factors included sex; ethnicity; age at NS diagnosis; age at SRNS onset; age at first rituximab; primary or secondary SRNS; histological subtype; prior number of immunosuppression use; duration of prior CNI use; rituximab dosing regimens; additional rituximab therapy; post-rituximab renin-angiotensin-aldosterone inhibitors (RAASi) use; post-rituximab concomitant immunosuppression use; 6-month remission status; as well as baseline CKD staging, kidney function, serum albumin levels, and proteinuria. Variables with P < 0.10 in univariate analysis were then included in a multivariate model. Multicollinearity in the multivariable model was evaluated using the variance inflation factor. The proportional hazards assumptions were tested using Schoenfeld residuals and visual inspection of scaled Schoenfeld plots. No imputation was performed for missing data. Missingness was minimal and quantified in Supplementary Table S2. Statistical analyses were performed using R version 4.4.3 with “survival”, “survminer” and “car” packages. Statistical significance was set at P < 0.05. Results Patient Population We included 151 children (males: n = 80, 53%; median age at SRNS onset: 6.8, interquartile range [IQR]: 3.7–9.9) years in the final analysis (Figure 1). All patients underwent kidney biopsy, and FSGS was the most common histological diagnosis found in 94 subjects (62%). Eighty-one (54%) and 70 (46%) children had primary and secondary SRNS, respectively. In primary SRNS, SR was established at initial presentation at a median age of 6.1 (IQR: 2.8–9.4) years. In contrast, children with secondary SRNS were first diagnosed with NS at a median age of 3.6 (IQR: 2.5–6.8) years and subsequently developed SR at a median age of 7.7 (IQR: 4.5–11.8) years.Figure 1Study flow chart. Study flow chart. Genetic tests were performed in 61 of 81 (75%) and 21 of 70 (30%) of patients with primary and secondary SRNS, respectively, with no genetic variant identified. Specifically, among the 61 children with primary SRNS, 52 (85%) underwent genetic test using either next-generation or whole-exome sequencing. There were no extrarenal manifestations indicative of syndromal or genetic diagnosis. Prior to rituximab treatment, CNIs (tacrolimus, n = 76; cyclosporine A, n = 75) were prescribed for 0–3, 3–6, 6–12 and >12 months in 32 (21%), 40 (26%), 40 (26%), and 39 (26%) children, respectively. Fifty-eight (38%) and 14 (9%) patients were treated with mycophenolate mofetil and cyclophosphamide. RAASi were used in 132 patients (87%). All patients were in nonremission (NR) when rituximab was initiated at the median of 8.4 (IQR: 4.3–18.1) months after the diagnosis of SRNS was established. At baseline, before rituximab therapy, 73 subjects (48%) had normal kidney function, and 47 (31%) and 31 (21%) had CKD stage 2 and 3, respectively. The median eGFR was 87.5 (IQR: 64.5–117.3) ml/min per 1.73 m2, UPCR was 3.9 (IQR: 2.6–10.2) mg/mg and serum albumin was 21.0 (IQR: 18.0-25.7) g/l. After rituximab, 129 patients continued with RAASi. Corticosteroids were the most common concurrent immunosuppression (n = 109, 72%), followed by tacrolimus (n = 84, 56%), mycophenolate mofetil (n = 50, 33%) and cyclosporine A (n = 39, 25%). In Table 1 and Supplementary Table S2, we summarize baseline demographics and treatment details. In Supplementary Tables S3 to S5, we present rituximab dosing regimen and therapeutic drug monitoring of CNIs. In Supplementary Figure S1 and Table S6, we report on immunosuppression use before and after rituximab.Table 1Treatment details pertaining to rituximab therapy and concomitant immunosuppression in 151 children with steroid-reisitant nephrotitc syndrome treated with rituximabTotalNo KF or deathKF or deathDifference (95% CI)N = 151N = 123N = 28Rituximab dosing/course375 mg/m223 (15.2%)19 (15.4%)4 (14.3%)1.2% (−14.4% to 16.8%)750 mg/m283 (55.0%)64 (52.0%)19 (67.9%)−15.8% (−37.4% to 5.8%)1125–1500 mg/m245 (29.8%)40 (32.5%)5 (17.9%)14.7% (−4.0% to 33.3%)Total B-cell depletion achieved after rituximab administrationa150 (99.3%)123 (100%)27 (96.4%)3.6% (−5.5% to 12.6%)Concomitant immunosuppression146 (96.7%)119 (96.7%)27 (96.4%)0.3% (−7.6% to 8.2%)Calcineurin inhibitors115 (76.1%)94 (76.4%)21 (75.0%)1.4% (−17.7% to 20.6%)Concomitant RAAS inhibition129 (85.4%)109 (88.6%)20 (71.4%)17.2% (−2.7% to 37.0%)Angiotensin converting enzyme inhibitor122 (80.8%)103 (83.7%)19 (67.9%)15.9% (−4.8% to 36.6%)Angiotensin receptor blocker34 (22.5%)29 (23.6%)5 (17.9%)5.7% (−12.5% to 24.0%)Additional rituximab therapyb35 (23.2%)32 (26.0%)3 (10.7%)15.3% (−0.7% to 31.3%)Remission status at 6 mosComplete Remission27 (17.9%)27 (22.0%)0 (0.0%)22% (12.4% to 31.5%)Partial Remission33 (21.9%)32 (26.0%)1 (3.6%)22.4% (9.9% to 35.0%)Nonremission91 (60.3%)64 (52.0%)27 (96.4%)−44.4% (−57.8% to −31.0%)Observation periods from rituximab, mos29.1 (16.7–62.9)27.9 (16.4–58.5)21.3 (12.3–36.9)6.0 (−0.3 to 15.0)CI. confidence interval; KF, kidney failure; RAAS, renin-angiotensin-aldosterone.aTotal B-cell depletion was monitored between 1 and 4 wks after the last infusion of the course of rituximab therapy.bAdditional rituximab therapy referred to the second course of rituximab therapy administered ≥3 mos after an initial course of rituximab. Table 1Treatment details pertaining to rituximab therapy and concomitant immunosuppression in 151 children with steroid-reisitant nephrotitc syndrome treated with rituximabTotalNo KF or deathKF or deathDifference (95% CI)N = 151N = 123N = 28Rituximab dosing/course375 mg/m223 (15.2%)19 (15.4%)4 (14.3%)1.2% (−14.4% to 16.8%)750 mg/m283 (55.0%)64 (52.0%)19 (67.9%)−15.8% (−37.4% to 5.8%)1125–1500 mg/m245 (29.8%)40 (32.5%)5 (17.9%)14.7% (−4.0% to 33.3%)Total B-cell depletion achieved after rituximab administrationa150 (99.3%)123 (100%)27 (96.4%)3.6% (−5.5% to 12.6%)Concomitant immunosuppression146 (96.7%)119 (96.7%)27 (96.4%)0.3% (−7.6% to 8.2%)Calcineurin inhibitors115 (76.1%)94 (76.4%)21 (75.0%)1.4% (−17.7% to 20.6%)Concomitant RAAS inhibition129 (85.4%)109 (88.6%)20 (71.4%)17.2% (−2.7% to 37.0%)Angiotensin converting enzyme inhibitor122 (80.8%)103 (83.7%)19 (67.9%)15.9% (−4.8% to 36.6%)Angiotensin receptor blocker34 (22.5%)29 (23.6%)5 (17.9%)5.7% (−12.5% to 24.0%)Additional rituximab therapyb35 (23.2%)32 (26.0%)3 (10.7%)15.3% (−0.7% to 31.3%)Remission status at 6 mosComplete Remission27 (17.9%)27 (22.0%)0 (0.0%)22% (12.4% to 31.5%)Partial Remission33 (21.9%)32 (26.0%)1 (3.6%)22.4% (9.9% to 35.0%)Nonremission91 (60.3%)64 (52.0%)27 (96.4%)−44.4% (−57.8% to −31.0%)Observation periods from rituximab, mos29.1 (16.7–62.9)27.9 (16.4–58.5)21.3 (12.3–36.9)6.0 (−0.3 to 15.0)CI. confidence interval; KF, kidney failure; RAAS, renin-angiotensin-aldosterone.aTotal B-cell depletion was monitored between 1 and 4 wks after the last infusion of the course of rituximab therapy.bAdditional rituximab therapy referred to the second course of rituximab therapy administered ≥3 mos after an initial course of rituximab. Treatment details pertaining to rituximab therapy and concomitant immunosuppression in 151 children with steroid-reisitant nephrotitc syndrome treated with rituximab CI. confidence interval; KF, kidney failure; RAAS, renin-angiotensin-aldosterone. Total B-cell depletion was monitored between 1 and 4 wks after the last infusion of the course of rituximab therapy. Additional rituximab therapy referred to the second course of rituximab therapy administered ≥3 mos after an initial course of rituximab. At 6 months after first rituximab administration, 60 children (40%) attained CR or PR (CR, n = 27, 18%; PR, n = 33, 22%) and 91 patients (60%) were in NR. The 6-month CR or PR rate was significantly higher in secondary SRNS (n = 40/70, 57.1%) compared with primary SRNS (n = 20/81, 24.7%; P < 0.001). Consistent with our previous findings,6 children treated with CNIs for <6 months before rituximab had a higher 6-month CR or PR rate (n = 35/72, 48.6%) than those who were CNI-resistant (≥ 6 months) (n = 25/79, 31.6%; P = 0.045). The 6-month remission status was comparable between tacrolimus and cyclosporine A (38.1% vs. 41.3%). The CR or PR rates were not different with or without concurrent CNI use post-rituximab (40.8% vs. 41.9%) or between ethnicities (Supplementary Table S7). The median observation periods from SRNS onset and rituximab to the last follow-up were 41.5 (IQR: 28.1–96.6) months and 29.1 (IQR: 16.7-62.9) months. Thirty-five of the 151 (23.2%) children received an additional course of rituximab for a median of 11.2 (IQR: 7.0–16.4) months) following first administration (3–6 months: n = 6; 6–12 months: n = 13; and > 12 months: n = 16 patients). The main indication for redosing was remission induction in 29 children, including those with NR (n = 10), PR (n = 7), or relapse after an initial CR or PR (n = 12). In the remaining 6 patients, rituximab was readministered for relapse prevention. Among the 10 children with NR, 6 (60%) achieved CR or PR following the second course of rituximab; whereas 3 (43%) of the 7 children with PR achieved CR. Nine of the 12 children who relapsed achieved CR or PR after retreatment with rituximab. Primary Outcome: Kidney Survival Twenty-eight of the 151 children (19%) developed kidney failure and started dialysis over a median time of 21.3 (IQR: 11.9–37.0) months. No patients received preemptive kidney transplant. Twenty-two of these 28 patients (78.6%) underwent genetic testing with no pathogenic variants identified (primary SRNS: 16/18, 89%; secondary SRNS: 6/10, 60%). Two deaths were reported after kidney failure and dialysis initiation. The overall kidney survival of the cohort was 82.7% (95% CI: 75.8%–90.2%), 75.8% (95% CI: 67%–85.7%) and 72.3% (95% CI: 62.1%–84.3%) at 3, 5, and 7 years post-rituximab (Figure 2), respectively. Eleven patients eventually received a kidney transplant, and 3 had posttransplant recurrence. Among the 140 patients who did not receive a kidney transplant, the median eGFR was 82 (IQR: 45–117) ml/min per 1.73 m2 and UPCR was 1 (IQR: 0.2-2.9) mg/mg at the last follow-up. In Table 1 and Supplementary Table S2, we compare children with or without progression to kidney failure or death.Figure 2Kidney survival of the 151 children with steroid-resistant nephrotic syndrome treated with rituximab. Kidney survival of the 151 children with steroid-resistant nephrotic syndrome treated with rituximab. Of the 28 subjects who developed kidney failure, 27 children were in NR and 1 in PR at 6 months. One nonresponding patient developed kidney failure within the first 6 months after rituximab. Significantly more children who reached kidney failure were in NR at 6 months (96% vs. 52%; P < 0.001). NR at 6 months was associated with inferior kidney survival than CR or PR (log-rank P < 0.0001; Figure 3a). The 1-, 3-, 5-, and 7-year kidney survival rates were 84.4% (95% CI: 76.9%–92.6%), 64.1% (95% CI: 52.6%–78.2%), 61.2% (95% CI: 49.2%–76.1%) and 57.1% (95% CI: 44.2%–73.8%) in the NR group, whereas kidney survival remained 100% (95% CI: 100%–100%) at all corresponding time-points among those who achieved CR or PR by 6 months.Figure 3Kidney survival stratified by (a) remission status at 6 months and (b) staging of CKD before rituximab therapy, using landmark analysis with time 0 defined as 6 months from rituximab administration. CKD, chronic kidney disease. Kidney survival stratified by (a) remission status at 6 months and (b) staging of CKD before rituximab therapy, using landmark analysis with time 0 defined as 6 months from rituximab administration. CKD, chronic kidney disease. Children who developed kidney failure or who died had a lower median eGFR at the time of rituximab (64.0 vs. 91.5 ml/min per 1.73 m2; P < 0.001). Kaplan-Meier analysis showed that higher CKD staging at the time of rituximab was associated with inferior kidney survival (log-rank test, P < 0.0001; Figure 3b). The 3-, 5-, and 7-year kidney survival rates were 94.2% (95% CI: 87.9%–100%), 91.7% (95% CI: 84%–100%) and 86.3% (95% CI: 74.4%-100%) in children with normal kidney function (CKD stage 1); 79.2% (95% CI: 66.1%–94.9%), 70.4% (95% CI: 52.5%–94.4%) and 70.4% (95% CI: 52.5%–94.4%) with CKD stage 2; and 60.2% (95% CI: 45.6%–84.6%), 35.5% (95% CI: 14.3%–88.1%) and 35.5% (95% CI: 14.3%–88.1%) with CKD stage 3. The kidney failure or death group also showed a higher baseline UPCR (8.3 vs. 3.6 mg/mg; P = 0.03) and lower serum albumin (19.0 vs. 22.0 g/l; P = 0.046). The receiver operating characteristics curve analysis identified cutoffs of UPCR ≥ 7.5 mg/mg (area under the curve = 0.633; sensitivity = 53.6% and specificity = 74%; P = 0.03) and albumin < 19.4 g/l (area under the curve = 0.621; sensitivity = 62.6% and specificity = 57.1%; P = 0.04) to predict kidney failure (Figure 4a and b).Figure 4Kidney survival stratified by (a) UPCR and (b) serum albumin levels. Time 0 was defined as the time of rituximab administration. The cutoff values for UPCR and serum albumin were determined using receiver operating characteristic curve analysis, based on Youden index. UPCR, urine protein-to-creatinine ratio. Kidney survival stratified by (a) UPCR and (b) serum albumin levels. Time 0 was defined as the time of rituximab administration. The cutoff values for UPCR and serum albumin were determined using receiver operating characteristic curve analysis, based on Youden index. UPCR, urine protein-to-creatinine ratio. Regarding histological type, FSGS was more prevalent in the kidney failure or death group (75% vs. 59%), more children with preserved kidney function had minimal change disease (MCD) on histology (33% vs. 11%) (P = 0.04). Kaplan-Meier analysis showed FSGS was associated with inferior kidney survival (log-rank P = 0.03; Figure 5a).Figure 5Kidney survival stratified by (a) histology and (b) additional course of RTX therapy. Time 0 was defined as the time of rituximab administration. For histology, others included 8 cases of mesangioproliferative glomerulonephritis, 3 cases of IgM nephropathy and 2 cases of C1q nephropathy. Additional rituximab therapy referred to the second course of rituximab therapy administered ≥3 months after an initial course of rituximab. RTX, rituximab. Kidney survival stratified by (a) histology and (b) additional course of RTX therapy. Time 0 was defined as the time of rituximab administration. For histology, others included 8 cases of mesangioproliferative glomerulonephritis, 3 cases of IgM nephropathy and 2 cases of C1q nephropathy. Additional rituximab therapy referred to the second course of rituximab therapy administered ≥3 months after an initial course of rituximab. RTX, rituximab. Kaplan-Meier analysis further demonstrated a different kidney survival among ethnicities (log rank P = 0.02; Supplementary Figure S2 and Table S7). Second course of rituximab was associated with superior kidney survival (Figure 5b). Other factors, including sex, type of SRNS (primary vs. secondary), prior mycophenolate mofetil use, duration of prior CNI treatment, and concomitant RAASi use were not associated with kidney survival (Supplementary Figures S3–S7). Predictors for Kidney Survival We performed a multivariable time-dependent Cox regression, which demonstrated that, before rituximab therapy, CKD stage 2 (HRadj: 2.7, 95% CI: 1.5–4.9; P < 0.001) and lower serum albumin levels (HRadj: 0.88, 95% CI: 0.81–0.95; P < 0.001), FSGS (HRadj 3.4, 95%CI 1.1-10.0, P = 0.03), and other histology (HRadj: 6.9, 95% CI: 1.7–27.7; P = 0.01) predicted inferior kidney survival. NR at 6 months (HRadj: 14.1, 95% CI: 2.2–92.5, P = 0.01) was a significant predictor of inferior kidney survival (Table 2 and Supplementary Table S8).Table 2Relationship with kidney failure and death using multivariable, time-dependent Cox proportional hazards modelsMultivariableUpper95% CIP-valuesHRadjLower95% CIEthnicity East Asian1 [REF] White1.430.603.42NS South Asian0.850.292.47NS Others3.530.6020.88NSHistology Minimal change disease FSGS3.351.1210.030.03 Others6.921.7327.730.006CKD stage prior to rituximab Normal kidney function (CKD stage 1) CKD stage 22.741.524.92< 0.001 CKD stage 33.190.8611.850.08Serum albumin prior to rituximab, per g/l0.880.810.95< 0.001Additional rituximab therapy0.290.081.03NSRemission status at 6 months14.142.1692.470.01CI, confidence interval; CKD, chronic kidney disease; CNI, calcineurin inhibitors; FSGS, focal segmental glomerulosclerosis; HRadj, adjusted hazard ratio; NS, nephrotic syndrome; RAAS: renin-angiotensin-aldosterone; SRNS, steroid-resistant NS; UPCR, urinary protein-to-creatinine ratio.Other factors considered in the selection process for multivariable model included sex, primary vs. secondary SRNS, age at NS diagnosis, age at SRNS onset, age at rituximab therapy, time between SRNS onset and rituximab administration, UPCR before rituximab, duration of CNI use before rituximab, rituximab dosing/course, number of concomitant immunosuppression post-rituximab, and concomitant RAAS inhibition post-rituximab. The full univariable and multivariable models are presented in Supplementary Table S8. Table 2Relationship with kidney failure and death using multivariable, time-dependent Cox proportional hazards modelsMultivariableUpper95% CIP-valuesHRadjLower95% CIEthnicity East Asian1 [REF] White1.430.603.42NS South Asian0.850.292.47NS Others3.530.6020.88NSHistology Minimal change disease FSGS3.351.1210.030.03 Others6.921.7327.730.006CKD stage prior to rituximab Normal kidney function (CKD stage 1) CKD stage 22.741.524.92< 0.001 CKD stage 33.190.8611.850.08Serum albumin prior to rituximab, per g/l0.880.810.95< 0.001Additional rituximab therapy0.290.081.03NSRemission status at 6 months14.142.1692.470.01CI, confidence interval; CKD, chronic kidney disease; CNI, calcineurin inhibitors; FSGS, focal segmental glomerulosclerosis; HRadj, adjusted hazard ratio; NS, nephrotic syndrome; RAAS: renin-angiotensin-aldosterone; SRNS, steroid-resistant NS; UPCR, urinary protein-to-creatinine ratio.Other factors considered in the selection process for multivariable model included sex, primary vs. secondary SRNS, age at NS diagnosis, age at SRNS onset, age at rituximab therapy, time between SRNS onset and rituximab administration, UPCR before rituximab, duration of CNI use before rituximab, rituximab dosing/course, number of concomitant immunosuppression post-rituximab, and concomitant RAAS inhibition post-rituximab. The full univariable and multivariable models are presented in Supplementary Table S8. Relationship with kidney failure and death using multivariable, time-dependent Cox proportional hazards models CI, confidence interval; CKD, chronic kidney disease; CNI, calcineurin inhibitors; FSGS, focal segmental glomerulosclerosis; HRadj, adjusted hazard ratio; NS, nephrotic syndrome; RAAS: renin-angiotensin-aldosterone; SRNS, steroid-resistant NS; UPCR, urinary protein-to-creatinine ratio. Other factors considered in the selection process for multivariable model included sex, primary vs. secondary SRNS, age at NS diagnosis, age at SRNS onset, age at rituximab therapy, time between SRNS onset and rituximab administration, UPCR before rituximab, duration of CNI use before rituximab, rituximab dosing/course, number of concomitant immunosuppression post-rituximab, and concomitant RAAS inhibition post-rituximab. The full univariable and multivariable models are presented in Supplementary Table S8. Discussion Our real-world data derives from a multiethnic pediatric cohort that represents the most refractory patients with presumably immune-mediated SRNS. We demonstrated that medium-term kidney survival following add-on rituximab therapy was 83% at 3 years and 72% at 7 years. In addition to nonresponse to rituximab, we identified preexisting CKD, lower baseline serum albumin levels, and non-MCD histological diagnoses as significant predictors of kidney failure and/or death. Our cohort is unique in comprising children with presumed immune-mediated SRNS who received both CNIs and rituximab. Males were slightly more prevalent in our cohort, with a lower male-to-female ratio than other reported cohorts (55%–65%).16,17 Nonetheless, sex was not found to be an independent predictor for both treatment response and long-term kidney survivals.8,16 Our cohort represents a high-risk population unresponsive to corticosteroids and multiple immunosuppressive agents. In the largest PodoNet registry, only two-thirds of children received CNIs and 1% received rituximab.8 The UK RaDaR Study reported 22% rituximab use among 190 genetic-test–negative children with SRNS,7 whereas a Japanese series of 21 children included only 6 treated with rituximab.10 Other large pediatric cohorts were treated exclusively with cyclophosphamide and/or CNIs.9,12, 13, 14 Moreover, recognizing that primary SRNS encompasses both genetic and immune-mediated forms, whereas secondary SRNS is generally immune-mediated,3,18,19 we minimized confounding by excluding patients with known genetic variants and histological features of genetic podocytopathies, such as diffuse mesangial sclerosis or Alport syndrome. In our cohort, 75% of children with primary SRNS underwent genetic testing, with >85% using next-generation or whole-exome sequencing, supporting a focus on presumed immune-mediated SRNS. Of note, almost 90% of children with primary SRNS and kidney failure were genetic-test–negative. We could thus with some certainty show that primary and secondary SRNS showed comparable kidney survival, contrasting with previous reports of superior outcomes in secondary SRNS, likely because of inclusion of genetic podocytopathies in the primary SRNS category.7 In our cohort, overall kidney survival was 83%, 76%, and 72% at 3, 5, and 7 years following rituximab therapy, whereas earlier studies reported 5-year survival of 72% to 75% in mixed SRNS cohorts, which were mostly not treated using rituximab.8,12,13 It should be noted that monogenic SRNS carries a poorer prognosis, with progression rates similar to immune-mediated, multidrug refractory SRNS.7 In this context, the 5-year survival for immune-mediated subgroup in those cohorts, though not explicitly reported, would be expected to exceed that observed in our study, which represents the most severe spectrum of immune-mediated SRNS. Our study underscores the importance of achieving remission in optimizing long-term patient and kidney outcomes. Findings from previous studies had consistently reported that nonresponse to immunosuppression predicted kidney failure.7, 8, 9,12,14 In this analysis, none of the children who achieved CR at 6 months died or progressed to kidney failure. Only 1 child with PR at 6 months developed kidney failure, and this occurred 10 years after rituximab therapy. By contrast, all the remaining 27 patients who reached kidney failure demonstrated a lack of treatment response to rituximab at 6 months. Although the 95% CI was wide, potentially reflecting the small sample size and limited number of events in the CR or PR group, the association between nonresponsiveness and adverse outcomes is compelling, and it is unlikely to be attributable to chance. The cumulative incidence of kidney failure among nonresponding patients reached 50% to 60% by 10 years, concurring with data from the PodoNet registry.8 These indicate that children who fail to achieve any remission, regardless of treatment, progress to kidney failure at a comparable rate. For patients who remain nonresponsive to the first rituximab by 6 months, our previous analysis showed that 64% who received additional rituximab therapy could achieve CR or PR.6 In the current study, redosing rituximab was associated with better kidney survival, although such effect was insignificant upon multivariable analysis. This finding should be interpreted with caution, because a proportion of patients were retreated for relapse prevention or to induce remission during a relapse episode. This subgroup responded to the first rituximab and is likely to represent a population with more favorable prognosis, which can cause confounding by indication. Alternative strategies, including obinutuzumab,20 adjunctive daratumumab,21 and/or enrolment into clinical trials,3 should be considered, with an aim of achieving at least PR to halt kidney failure progression. A single-arm Japanese study suggested that concomitant pulse corticosteroids following rituximab enhanced proteinuria reduction.22 We identified preexisting CKD before rituximab as a significant predictor of kidney survival, resulting in approximately a 3-fold increase in developing kidney failure. Previous studies reported inconsistent conclusions and primarily assessed the kidney function measured at NS onset,8,13 rather than before a specific intervention such as rituximab later in the disease course. This reflects the irreversible and progressive nature of chronic structural injuries such as nephron loss, glomerulosclerosis, and interstitial fibrosis, where kidney function continues to decline through mechanisms including hyperfiltration.23 These findings emphasize the importance of considering alternative therapeutic strategies early, including rituximab, in SRNS before CKD ensues. Our previous data indicated that early add-on rituximab therapy to CNIs was associated with a higher remission rate before CNI-resistance is established, although this may reflect a combined effect of rituximab and CNIs. Intensified immunosuppression should be accompanied by intensive CKD care to slow kidney failure progression,23 including vitamin D supplementation and optimization of blood pressure, as well as lipid and glucose control. In our cohort, RAASi was regarded an integral SRNS management by most pediatric nephrologists, with 85% of children receiving the medication concurrently after rituximab. RAAS inhibition is a well-established strategy for reno-protection in children and adults, and in this study RAAS inhibition was associated with improved kidney survival though not reaching statistical significance because of the small number of patients who did not receive RAASi. Although robust evidence has demonstrated the reno-protective effects of sodium-glucose cotransporter 2 inhibitors in adult patients with CKD, the medications are not licensed for pediatric use and available data is lacking. Lower serum albumin before rituximab predicted kidney failure, likely reflecting severity and duration of proteinuria. All children in our cohort underwent kidney biopsy, enabling a detailed assessment of the relationship between histology and kidney survival. Traditionally, childhood SRNS secondary to FSGS has been associated with a poorer prognosis, compared with MCD.8,9,12 Similar findings were observed in the current work, where FSGS and other non-MCD histological subtypes (IgM nephropathy, C1q nephropathy, and mesangioproliferative glomerulonephritis) are influential determinants of kidney outcomes. Interestingly, our earlier analysis, consistent with previous studies,7 failed to demonstrate histology as a predictor of response to rituximab.24 Furthermore, the duration of prior calcineurin inhibition, whether fulfilling the 6-month criterion for treatment resistance, did not impact kidney survival. Similarly, post-rituximab CNI use was not associated with NS remission or kidney survival. Antinephrin or antislit autoantibodies have been implicated in SRNS,25,26 though in smaller proportions than in steroid-sensitive NS, and predict responsiveness to immunosuppression. Large-scale collaborative studies are urgently needed to define their prognostic value in guiding treatment selection and informing long-term kidney outcomes. A major strength of the present study is the inclusion of a well-characterized, multiethnic pediatric cohort with minimal missing data. Nevertheless, several limitations should be acknowledged. First, the retrospective design introduced potential selection and reporting bias, although this was mitigated by stringent inclusion criteria. Variability in sample collection and quality, such as the timing and adherence to first-morning void collection, may introduce bias into our results. Second, the observation period after rituximab was relatively short, restricting kidney survival analysis to short- to medium-term outcomes. Third, though we excluded patients with known genetic podocytopathies, genetic testing was performed according to institutional practice and test availability. This raises the possibility of including children with undetected genetic disease. Fourth, there were considerable variations in rituximab dosing regimens (375–1500 mg/m2), immunosuppression protocol and B-cell monitoring practice across centers, which could have influenced response rates and consequently overall kidney survival of the cohort. We tried to adjust this by incorporating center effect and rituximab dosing regimens in our multivariable regression model for kidney survival. Our previous findings did not find a significant association between rituximab dosing and remission outcomes.24 Fifth, although CKD stage at rituximab initiation predicted kidney failure, this association may reflect underlying disease severity rather than a modifiable treatment window and causality cannot be inferred. Lastly, Black patients were underrepresented in this cohort. This group may carry high risk APOL1 allele, presenting with more severe disease and histological diagnoses other than MCD.27 This underrepresentation may limit the generalizability of our findings. Despite these limitations, our study provides important real-world outcome data of a rare disease to inform patient counselling and management. In conclusion, we have demonstrated that medium-term kidney survival among children with the most severe spectrum of SRNS treated with add-on rituximab ranges between 70% and 80%. In addition to treatment nonresponse, preexisting CKD predicts inferior kidney survival, underscoring the need to consider therapies such as rituximab early, before CKD develops. Disclosure HGK received grants from ChongKunDang, Handok, and Kyowa Kirin; consulting fees from Handok; honoraria from Alexion and Handok; and participated in advisory board for Bayer Pharmaceuticals. MV had consultancy agreements with Purespring Therapeutics, Bayer Pharmaceuticals, Travere Therapeutics, Roche Pharmaceuticals, and Chinook Pharmaceuticals; and participated in advisory board for Roche Pharmaceuticals. Other the othe authors declared no competing interests.