This investigator-initiated, open-label, multicenter, prospective cohort study was conducted at 10 sites in coastal cities of Jiangsu Province, eastern China. The study design has been reported previously (11). The study was conducted and reported in accordance with the STROBE checklist.

Beginning November 1^st 2019, patients presenting with nephrotic syndrome at the 10 participating nephrology centers were screened for eligibility. Inclusion criteria were as follows: (1) biopsy-proven membranous nephropathy (MN) at first diagnosis with moderate or high risk of renal progression, or refractory MN; (2) age 18–75 years with proteinuria >3.5 g per 24 h and serum albumin <30 g/L; and (3) CD19⁺ B lymphocyte count >5/mm³. Exclusion criteria were: (1) estimated glomerular filtration rate (eGFR) <30 mL/min/1.73 m²; (2) infant or childhood-onset nephrotic syndrome; (3) secondary MN; (4) abnormal liver function (greater than two times the upper limit of normal); (5) pregnancy or breastfeeding; (6) active infectious diseases, such as chronic hepatitis B, hepatitis C, AIDS, or tuberculosis; (7) severe impaired immune response, such as hypoimmunoglobulinemia (IgG <4 mg/dL), CD4 cell count <200/mm³, or CD19⁺ B lymphocyte count < 5/mm³, (8) major cardiovascular or cerebrovascular events (e.g., myocardial infarction, heart failure, cerebral hemorrhage) within the past 6 months; (9) systemic immunosuppressant use for more than 2 weeks within 12 weeks before screening, with inability to discontinue or taper. Immunosuppressants (cyclophosphamide, mycophenolate mofetil, azathioprine, or Tripterygium wilfordii) were prohibited for at least 3 months before enrollment. For refractory MN previously treated with calcineurin inhibitors (CNIs), tacrolimus or cyclosporine could be continued at tapering doses with serum concentrations maintained at 4–8 ng/mL (tacrolimus) or 100–150 ng/mL (cyclosporine) at study entry. Steroid doses did not exceed 20 mg/day. (10) Allergy to rituximab or any excipient in the formulation. All patients were followed for 12 months after hospital discharge.

The study was approved by the ethics committee of the First Affiliated Hospital of Nanjing Medical University (2019-SR-452.A1) and the ethics committees of the other 9 sites. The trial was registered in the Chinese Clinical Trial Registry (ClinicalTrials.gov identifier ChiCTR1900026382). Informed consent was obtained for treatment, follow-up, and tissue and blood sampling.

Refractory membranous nephropathy (12, 13) was defined as the absence of clinical and/or immunological remission (i.e., antibody titer below the detection threshold by ELISA or a negative indirect immunofluorescence assay) after a course of treatment with corticosteroids and a calcineurin inhibitor (CNI).

Moderate or high risk of renal progression was defined according to KDIGO guidelines: Moderate risk: Normal eGFR; proteinuria > 3.5g/d and not decreased >50% after 6 months of conservative therapy with an ACEi/ARB; and not fulfilling high-risk criteria. High risk: eGFR <60 ml/min/1.73 m² and/or proteinuria >8g/d for >6 months; or normal eGFR, proteinuria >3.5 g/day and not decreased by >50% after 6 months of conservative therapy with ACEi/ARB, plus at least one of the following: serum albumin <25g/l, anti-PLA2R antibody >50 RU/mL, urinary IgG>1ug/min, urinary α1-microglobulin >40 µg/min, urinary β2-microglobulin >250 mg/day, or selectivity index >0.20 (14).

In view of personal preference, patients were assigned into these two groups: individualized therapy and standardized therapy according to their willingness.

Strategy of individualized low-dose (B cell- and anti-PLA2R-targeted) therapy was conducted according to the treatment schedule of IMN patients assigned in our previous study (11). The initial dose of rituximab depended on the body surface area (BSA), and the minimum dose was 150 mg/m² BSA. Considering that the half-life of rituximab (RTX) is approximately 3 weeks (15), the second injection was scheduled 2 weeks later to maintain stable therapeutic drug concentrations. After that, follow-up was conducted at an interval of 2 months. At each follow-up, if >5 B cells per mm³ were observed, (1) if the PLA2R antibody decreased or was negative, 100 mg was administered; (2) if the PLA2R antibody was unchanged, 75 mg/m² was given; (3) if the PLA2R antibody was higher than before, 150 mg/m² was given. If <5 B cells per mm³ were observed, (1) if the PLA2R antibody decreased or was negative, no additional course; (2) if the PLA2R antibody was unchanged, 100 mg was administered; (3) if the PLA2R antibody was higher than before, 75 mg/m² was given.

In brief, our initial dose during the first month was determined by peripheral B cells and body surface area. Follow-up every 2 months thereafter was planned. The subsequent dosage depended on the anti-PLA2R antibody titer level and B cell count.

All patients received optimal supportive care, including renin–angiotensin system blockers and blood pressure management. For refractory MN, the doses of calcineurin inhibitors could be maintained or tapered with serum trough concentrations within 4–8 ng/mL (tacrolimus) or 100–150 ng/mL (cyclosporine) after entering the study. Steroids did not exceed 20 mg daily. Patients were followed for 12 months after discharge from the hospital.

Sixty-three patients per group would provide 80% power to detect noninferiority regarding complete remission (CR) or partial remission (PR) at 12 months at a one-sided significance level of 0.025 (equivalent to a two-sided significance level of 0.05) and a noninferiority margin of 15 percentage points on an absolute risk difference scale, assuming that 55% of patients in the individualized rituximab group and 45% of those in the standard therapy group achieved CR or PR at 12 months. Considering a 10% dropout rate, at least 70 patients per group were enrolled.

The primary clinical outcome was the composite of complete or partial remission at 12 months. Secondary outcomes were time to remission, progression to end-stage kidney disease (ESKD), anti-PLA2R levels, proteinuria, estimated glomerular filtration rate, and adverse events.

The following definitions were used: complete or partial remission, proteinuria <0.3 g/24 h or <3.5g/24h and <50% of baseline, respectively, in at least two consecutive visits; relapse, recurrence of massive proteinuria >3.5 g/24 h and serum albumin <30g/L on two of three consecutive days; CD19⁺ B cell depletion, CD19⁺ B cell count <5/mm³. For safety evaluation, immediate infusion reactions (within 48 h) were recorded. Serious infections were defined as any infection requiring hospitalization and/or intravenous antibiotics or resulting in disability or death.

The analysis was performed using R software version 4.2. Continuous variables with normal distribution were expressed as mean ± standard deviation and analyzed using Student’s t-test. Nonnormally distributed continuous variables were expressed as median (interquartile range, IQR) and tested using the Wilcoxon rank-sum test. Categorical variables were presented as n (%), and the chi-squared test or Fisher’s exact test was used. Logistic regression was used to calculate risk difference (RD) and 95% confidence interval (CI) to compare remission rates between individualized therapy and standard therapy. Survival rates were estimated by Kaplan–Meier analysis, and survival curves were compared using the log-rank test.

We applied inverse probability of treatment weighting (IPTW) by computing stabilized weights inversely proportional to the probability of treatment assignment to control for potential confounders. Standardized differences were used to assess balance in characteristics between groups after IPTW, with an absolute value <0.10 considered negligible. The missing indicator method was applied for covariates with missing data. Covariates included MN clinical type, laboratory values, and medications. Sociodemographic characteristics at index date (age, sex, BMI, systolic/diastolic blood pressure, and duration of proteinuria), prior therapies, and therapies during follow-up were also included. A p-value <0.05 was considered statistically significant.

Among 204 patients referred to our 10 nephrology units from 1 November 2019 to 15 June 2023, 53 did not meet the inclusion criteria. The remaining 151 patients were assigned to each rituximab therapy group for the management of IMN. Five participants in the individualized group and six in the standard therapy group were excluded from statistical analysis because of loss to follow-up within 1 month after rituximab use, or due to infusion reactions at first administration leading to withdrawal after enrollment. Thus, they were not included in the overall effect evaluation. Over a median (interquartile range) follow-up of 12 (12, 16) months, a total of 140 patients (78 in the individualized therapy group and 62 in the standard therapy group) were available for statistical analysis (see flow chart in Figure 1 ).

The baseline parameters of the two groups are shown in Table 1 . The median age was 51 years (SD, 14.59) in the individualized group and 52 years (SD, 13.99) in the standard group (standardized difference, 0.059); 51.2% and 50.8% were men (standardized difference, 0.002), respectively; and 19% and 40% were initial rituximab users (standardized difference, 0.884), respectively ( Table 1 ). Baseline characteristics were similar between the individualized and standard groups; however, most patients (64.5%) in the standard therapy group received rituximab as initial therapy, according to clinical criteria for assessing the risk of progressive loss of kidney function in the KDIGO (14) guidelines (14). Gender, age, BMI, eGFR, serum albumin, urinary protein, anti-PLA2R, CD19⁺ B cells, and IgG were well balanced between groups before IPTW. Differences in previous immunosuppressant use (P <0.001), immunosuppressant therapy during follow-up (P <0.001), and duration of proteinuria before rituximab initiation (P = 0.016) in the standard therapy group were further balanced after IPTW (P >0.05) ( Table 1 ).

Primary outcome: remission at one year.

At 12 months of follow-up, 57 of 78 (73.1%) patients in the individualized therapy group and 40 of 62 (64.5%) in the standard therapy group achieved complete or partial remission (P <0.001). At 12 months, a significantly higher proportion of patients in individualized therapy (n=27; 34.6%) reached complete remission compared with standard therapy (n=11; 17.7%) (P <0.001) ( Table 2 ). The median (IQR) remission time in both groups was 6 (4, 12) months. Multivariate logistic regression was used to evaluate treatment effect with adjustment for potential confounding variables. The outcome described whether patients maintained PR or CR within 12 months. Therapy strategy was included in the model along with other covariates, including gender, age, BMI, first-line rituximab use, anti-PLA2R titer, 24-h urinary protein, and use of steroids or immunosuppressants. Stepwise regression screening was performed, with both inclusion and exclusion criteria set at 0.05. The final model covariates were gender, 24-h urinary protein, and first-line rituximab use. The adjusted risk difference of response and 95% confidence interval was 0.1 (–0.05 to 0.26); the lower end of the confidence interval was above –15 percentage points, and the one-sided p-value for noninferiority was 0.001, meeting the significance threshold of α = 0.025 ( Table 3 ). In unweighted cohorts, 73.1% of 78 individualized treatment patients and 66.1% of 62 standard treatment patients reached partial or complete remission (weighted risk ratio, 1.07 [95% CI, 0.92–1.25], p = 0.4). The weighted remission rates were 74.1% in the individualized group and 70.5% in the standard group (weighted risk ratio, 1.2 [95% CI, 0.71–2.07], p = 0.5) ( Table 4 ).

The log-rank test was used to compare cumulative event rates of partial or complete remission ( Figure 2A ), partial remission ( Figure 2B ), and complete remission ( Figure 2C ). The results indicated that cumulative event rates between the two therapy groups were not significantly different.

From baseline to 12 months, the median urinary protein decreased from 7.33 g/d (IQR 4.73, 10.69) to 0.95 (IQR 0.23, 3.68) g/d in individualized therapy, and from 6.13 g/day (IQR 4.44, 9.54) to 1.36 g/day (IQR 0.40, 3.90) in standard therapy ( Table 2 , Supplementary Figure 1A , P = 0.445). Albumin increased from 22.90 (IQR 19.10, 26.70) g/L to 38.90 (IQR 34.30, 42.10) g/L in individualized therapy, and from 22.75 (IQR 18.15, 26.25) g/L to 38.30 (IQR 32.05, 41.05) g/L in standard therapy ( Table 2 , Supplementary Figure 1B , P = 0.957). Renal function remained stable over time in both groups ( Supplementary Figure 1E ).

Within the cohort of 110 participants available for anti-PLA2R antibody measurement at baseline, titers exceeded the threshold of 14 RU/mL used to define antibody positivity. Anti-PLA2R levels decreased in both groups during follow-up at a similar rate ( Supplementary Figure 1C ). Among 89 patients with detectable anti-PLA2R at baseline, 54 achieved anti-PLA2R negativity (<14 RU/mL) within 12 months after rituximab administration. Of these 54 patients, 47 subsequently achieved the study endpoint, compared with only 19 of the 35 patients without antibody negativity (P <0.001).

Most patients in the individualized therapy group achieved depletion of circulating CD19⁺ lymphocytes within 2 months after the first rituximab infusion, whereas all patients in the standard therapy group achieved depletion. The median time to CD19⁺ cell depletion was 2 months (IQR 2, 6) in individualized therapy and 2 months (IQR 2, 2) in standard therapy. CD19⁺ cell counts remained depleted in most patients in individualized therapy during follow-up, whereas recovery was observed from 6 months onward in the standard group, with significantly higher counts at 12 months [6.42 (IQR 2.24, 23.10) vs. 16.10 (IQR 5.23, 100.00), P = 0.024] ( Supplementary Figure 1D ).

In individualized therapy, 78 patients underwent CD19⁺ and anti-PLA2R testing every 2 months. The median rituximab dose at 1 year was 800 mg (IQR 600, 1,100) per patient, with a total cost of RMB 16,227.5 (IQR 13,148, 23,536) per unit utility. In the standard therapy group, 32 patients received 1 g on days 0 and 15, and 30 patients received 375 mg/m² weekly for 3 to 4 weeks. The median rituximab dose at 1 year was 2,000 mg (IQR 2,000, 2,400) per patient, with a total cost of RMB 31,265 (IQR 31,265, 40,163.2) per unit utility, which was significantly higher than that in individualized therapy (P <0.001). Detailed data are shown in Table 2 .

Three patients in the individualized therapy group and two in the standard therapy group relapsed during follow-up. No statistical difference was found between groups (P = 0.814). One of the five patients who relapsed was antibody-negative, while the other four were antibody-positive. Of the five, two patients relapsed at the 1-year screening, probably due to re-emergence of anti-PLA2R autoantibodies at the 10-month visit. The other two relapsed without re-emergence of anti-PLA2R antibody or CD19⁺ B cells.

Seven patients in the individualized therapy group and 10 in the standardized therapy group experienced at least one side effect, including infusion-related reactions and infections. No cases of leukopenia were observed. Two patients in the individualized group and four patients in the standard group suffered from pneumonia requiring hospitalization and intravenous antibiotics (see Supplementary Table 1 , χ2 = 125.56, P<0.01). There were statistically significant differences in the frequency of adverse events between groups (χ2 = 89.167, P<0.01). No cancer diagnoses or deaths occurred during the trial ( Supplementary Table 1 ).