B-cell landscape remodelling following dual BAFF/APRIL inhibition with telitacicept in lupus nephritis: extrafollicular axis clearance and its association with clinical response

Abstract

Objective

The therapeutic efficacy of telitacicept, a novel dual BAFF/APRIL inhibitor in active lupus nephritis (LN) requires further characterisation. This prospective cohort study characterised the B-cell landscape changes in patients with LN receiving telitacicept with a small belimumab-treated reference group.

Methods

Patients with active LN received telitacicept (n=19) or belimumab (n=5) for 24 weeks alongside standard therapy. Peripheral B-cell subsets were analysed at baseline, week 4, 12 and 24 using full-spectrum flow cytometry. The primary endpoint was the proportion of patients achieving primary efficacy renal response (PERR) at week 24.

Results

After excluding 3 dropouts, 16 telitacicept and 5 belimumab were included in the final analysis. In the telitacicept group, 43.8% and 62.5% of patients achieved PERR at week 4 and week 24, respectively, showing a progressive increasing trend alongside the belimumab reference group. Telitacicept was associated with numerically greater reductions in 24-hour urine protein and anti-double-stranded DNA titres, alongside serological improvements relative to the belimumab arm. While both treatments contracted the naïve B-cell pool, telitacicept uniquely induced more profound and sustained depletion of the pathogenic extrafollicular axis. Specifically, plasmablasts and mature plasma cells were effectively eradicated in the telitacicept group, whereas these subsets remained largely refractory to belimumab. This comprehensive ablation resulted in a significant relative enrichment of switched memory B-cells, which negatively correlated with disease activity, suggesting a passive reshaping of the residual B-cell pool rather than an active immune reset. Notably, early reductions in naïve B-cells (area under the curve (AUC)=0.909), mature plasma cells (AUC=0.964) and plasmablasts (AUC=0.836) were identified as early immunological correlates of sustained renal response.

Conclusion

Telitacicept was associated with numerically greater clinical and immunological remission by targeting the naïve-extrafollicular axis that evades single BAFF inhibition. Early depletion of plasma cells and naïve B-cells serves as promising molecular indicators associated with achieving sustained renal remission.

Trial registration number

NCT05588830.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• While BAFF inhibition with belimumab improves outcomes in lupus nephritis (LN), many patients remain refractory. The prospective mechanistic impact of telitacicept, a dual BAFF/APRIL inhibition, on B-cell landscape remodelling and its association with renal recovery remains insufficiently characterised.

WHAT THIS STUDY ADDS

• Telitacicept was associated with favourable numerical improvements in renal response and autoantibody clearance. Through dual BAFF/APRIL inhibition, it achieves a more profound depletion of the naïve-extrafollicular axis (including double-negative 2 B-cells and plasmablasts) that often escapes single BAFF inhibition.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Early on-treatment reductions in naïve B-cells and the plasma cell lineage serve as promising immunological correlates associated with sustained renal response, supporting the potential for personalised B-cell monitoring to optimise therapeutic strategies in LN.

Introduction

SLE is a heterogeneous autoimmune disease characterised by loss of immune tolerance and autoantibody production, which can lead to multiorgan inflammation.^{1 2} Lupus nephritis (LN) develops in up to 60% of patients with SLE and represents a leading cause of morbidity and mortality.^{3 4} Although conventional immunosuppressive therapies have improved outcomes, a significant number of patients experience refractory or relapsing disease, underscoring the need for more effective treatment strategies.⁵

B-cells play a central role in LN pathogenesis through autoantibody production, antigen presentation, proinflammatory cytokine secretion and formation of immune complexes. ^6–8Aberrations in specific B-cell subsets, particularly double-negative (DN) B-cells, are closely linked to disease activity,⁹ providing a rationale for B-cell-targeted therapies. Belimumab (BLM), a monoclonal antibody that inhibits the B-cell survival factor BAFF (B lymphocyte stimulator), was the first biological approved for SLE and LN.^{10 11} It modulates B-cell subsets, primarily reducing naïve and transitional B-cells with a secondary decline in memory B-cells and plasma cells.^12–14 Nonetheless, delayed or incomplete responses in a considerable proportion of patients highlight the demand for alternative agents.^{15 16} Recent studies have highlighted the extrafollicular pathway as a major driver of severe autoimmune flares. In this axis, hyperactivated DN2 B-cells bypass the classical germinal centre reaction to rapidly differentiate into pathogenic, autoantibody-secreting plasmablasts and plasma cells.^{17 18}

Given that extrafollicular-derived plasma cells often exhibit diminished sensitivity to BAFF inhibition alone, dual targeting of both BAFF and APRIL is required to effectively neutralise this pathogenic axis. Telitacicept (TAI), a novel TACI-Fc approved in China for SLE in 2021, simultaneously inhibits both cytokines, enabling more profound suppression of autoreactive B-cells and plasma cells to promote earlier and sustained clinical remission.^{19 20} Phase 2 and 3 clinical trials, along with real-world studies, have demonstrated its efficacy in SLE, including benefits in patients with LN such as rapid reduction of proteinuria and steroid-sparing effects.^21–24 Recent comparative analyses suggest that telitacicept may lead to superior and earlier clinical responses compared with belimumab.^{25 26}

However, existing comparative studies are predominantly retrospective and lack prospective, mechanistic insights into the dynamic changes of B-cell subsets that could explain differential clinical efficacy. The impact of dual BAFF/APRIL inhibition versus BAFF inhibition alone on the B-cell landscape in patients with LN remains insufficiently characterised. Therefore, we conducted a prospective cohort study to characterise the clinical outcomes and longitudinal alterations in circulating B-cell subsets in patients with active LN receiving telitacicept. A small belimumab-treated cohort was included as a descriptive pharmacological reference to identify immunological correlates of response and provide a rational basis for optimising B-cell-targeted strategies in LN.

Materials and methods

Study design and participants

This was a multicentre, prospective, controlled cohort study. Although initially designed as a multicentre study across 11 clinical sites in China, practical constraints resulted in patient enrolment being ultimately conducted at two participating centres (see online supplemental table S1). The final participant cohort consisted of patients from these sites. The study was registered at ClinicalTrials.gov (NCT05588830).

We recruited patients aged 18–65 years who met the 1997 American College of Rheumatology (ACR) or the 2019 European Alliance of Associations for Rheumatology (EULAR)/ACR classification criteria for SLE,^{27 28} and had active LN. Key inclusion criteria were: (1) SLE Disease Activity Index 2000（SLEDAI-2K) score ≥8 and urine protein to creatinine ratio (UPCR) ≥0.5 mg/mg (or equivalent 24-hour urine protein), with or without active urinary sediment; (2) biopsy-proven Class III, IV (±V) or pure Class V LN within the preceding 6 months; (3) <25% reduction in proteinuria after a minimum of 3 months of standard therapy; and (4) positive ANA (≥1:80) and/or anti-double-stranded DNA (dsDNA) antibodies.

Exclusion criteria included: (1) severe hepatic or renal impairment (alanine aminotransferase/aspartate aminotransferase >2 times upper limit of normal, estimated glomerular filtration rate (eGFR) <30 mL/min/1.73 m² or serum creatinine ≥265 µmol/L); (2) requirement for renal replacement therapy or transplantation; (3) active central nervous system lupus or thrombotic microangiopathy; (4) codiagnosis of other systemic autoimmune diseases; (5) pregnancy or lactation; (6) immunodeficiency, uncontrolled severe infection or active/recurrent peptic ulcer; (7) prior treatment with any biological agent or mesenchymal stem cell infusion.

Treatment protocol

In addition to standard-of-care therapy (glucocorticoids, hydroxychloroquine and/or immunosuppressants such as mycophenolate mofetil or tacrolimus), enrolled patients received either telitacicept or belimumab for 24 weeks. To minimise confounding from background therapies, all patients had been receiving stable doses of standard-of-care immunosuppressants and glucocorticoids for at least 3 months prior to the initiation of biological agents. The treatment assignment was non-randomised and jointly determined by the physician and patient through a shared decision-making process that incorporated clinical considerations, drug accessibility, economic factors and patient preference. This led to unequal group sizes and baseline imbalances, notably the higher proteinuric burden in the belimumab group (predominantly Class V) and the growing economic viability of telitacicept in China following its recent approval. Telitacicept was administered subcutaneously at a dose of 160 mg weekly, whereas belimumab was administered intravenously at 10 mg/kg on days 0, 14 and 28, and then every 4 weeks. Doses of concomitant glucocorticoids and immunosuppressants could be tapered based on clinical response; however, any increase in dosage or the introduction of new immunosuppressive agents was considered a treatment failure and led to study withdrawal.

Clinical efficacy assessment

The primary efficacy endpoint was the proportion of patients achieving a primary efficacy renal response (PERR) at week 24. PERR was defined as UPCR ≤0.7 g/g with an eGFR no worse than 20% below the baseline value or ≥60 mL/min/1.73 m².²⁹ Secondary endpoints included the proportion achieving complete renal response (CRR; UPCR ≤0.5 g/g with eGFR normal or no worse than 10% below baseline),²⁹ as well as complete remission (CR), partial remission (PR) or no response (NR) based on predefined criteria for proteinuria, serum albumin and serum creatinine.³⁰

Clinical and laboratory parameters, including SLEDAI-2K score, 24-hour urine protein, serum creatinine, serum albumin, erythrocyte sedimentation rate (ESR), complement levels (C3, C4), immunoglobulins (IgG, IgA, IgM) and anti-dsDNA titres, were assessed at baseline and at weeks 4, 12 and 24. Changes in prednisone-equivalent dose and the occurrence of adverse events were recorded throughout the study.

Analysis of B-cell subsets

Anticoagulated blood was collected at baseline at all scheduled follow-up visits. Peripheral blood mononuclear cells were freshly isolated via Ficoll density gradient centrifugation for subsequent flow cytometry analysis. Cells were stained with 7-AAD (viability marker) and a cocktail of fluorochrome-conjugated mouse anti-human monoclonal antibodies—including CD3-BB700, CD19-APC-Cy7, CD20-PE/Cyanine7, CD11c-BV421, IgD-FITC, CD27-PE, CD38-BV650, CD138-BV711, CXCR5-PE/Dazzle 594 and CD21-APC—for 30 min at 4℃ in phosphate-buffered saline containing 0.5% bovine serum albumin. Data were acquired on a Cytek Aurora full spectrum cytometry system using SpectroFlo software and subsequently analysed with FlowJo V.10.8.1 (BD Biosciences). To minimise Poisson counting errors for rare events,³¹ downstream frequencies of B-cell subsets were only calculated for samples containing a minimum of 1000 viable CD3⁻CD19⁺ parent B-cell events. Absolute counts of B-cell subsets were calculated using a dual-platform approach: the percentage of each subset within the total CD19⁺ B-cell population was determined by flow cytometry and then applied to the absolute CD19⁺ B-cell count obtained from the concurrent clinical haematological analysis.

To rigorously exclude non-B-cell contaminants, a stringent gating strategy was employed (online supplemental figure S1). The total B-cell lineage was defined as CD3⁻CD19⁺ events. Within this compartment, antibody-secreting cells (ASCs) were strictly identified by the hallmark CD20⁻CD38⁺⁺CD27⁺⁺ phenotype to eliminate myeloid artefacts. Circulating plasmablasts and mature plasma cells were further distinguished as CD138⁻ and CD138⁺ subsets, respectively.¹⁸ Conversely, the mature B-cell pool was identified as CD20⁺CD38^dim/−. In accordance with recent consensus nomenclature avoiding the term ‘Age-associated B-cells (ABCs)’,³² pathogenic DN2 cells were explicitly gated as CD11c⁺CXCR5⁻CD21⁻ events exclusively from the mature DN (IgD⁻CD27⁻) pool.¹⁷ Gating boundaries for activation markers were validated using Fluorescence Minus One controls.

Statistical analysis

Given the small sample size and to avoid artificially reducing variance, no missing data imputation was performed; all analyses were strictly restricted to available cases. Continuous variables are presented as median with IQR and categorical variables as number (percentage). Non-parametric approaches were uniformly adopted. Between-group comparisons were performed using the Mann-Whitney U test for continuous variables and Fisher’s exact test for categorical variables. Longitudinal within-group changes from baseline were evaluated using the Wilcoxon matched-pairs signed-rank test. To rigorously control the false-positive risk associated with multiple testing, all resulting p values were adjusted using the Benjamini-Hochberg false discovery rate method, with an adjusted q-value <0.05 considered statistically significant.

For biomarker evaluation, receiver operating characteristic (ROC) curve analyses were conducted using percentage changes from baseline to early time points. To avoid mathematical fragility and overfitting in small cohorts, exact cut-off thresholds were omitted; instead, the area under the curve (AUC) alongside the 95% CI were reported. The optimal cut-off point was determined using the Youden index for reference purposes only. Given the exploratory nature of the biomarker analysis, the AUC values are presented without cross-validation and should be interpreted cautiously. These findings require independent validation. All statistical analyses were performed using GraphPad Prism (V.10.3.0).

Reporting guideline

This study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement.³³ The completed STROBE checklist is provided in online supplemental file 2.

Results

Baseline characteristics of patients with LN

From November 2022 to April 2025, 24 eligible patients with active LN were enrolled, with 19 assigned to telitacicept and 5 to belimumab. Three patients in the telitacicept group discontinued the study (2 lost to follow-up at week 4, 1 withdrew consent at baseline), resulting in 16 and 5 patients in the telitacicept and belimumab groups, respectively, completing the 24-week treatment course and comprising the full analysis set (figure 1).

Figure 1. Patient screening and enrolment. This flowchart illustrates the patient selection process. Following exclusion, a total of 21 patients were ultimately enrolled and received B-cell-targeted therapy, with 16 completing the study in the telitacicept group and 5 in the belimumab group.

Baseline demographic and clinical characteristics are summarised in table 1. All patients were female, with a median age of 35.0 (IQR 29.5–45.0) years in the telitacicept group and 36.0 (IQR 27.5–41.0) years in the belimumab group. Consistent with the real-world, non-randomised treatment allocation driven by shared physician–patient decision-making, several baseline parameters exhibited notable intergroup imbalances. Specifically, patients in the belimumab group presented with higher baseline 24-hour proteinuria (median 4.79 vs 2.17 g/day) and a higher proportion of Class V LN (60% vs 12.5%), whereas the telitacicept group tended toward higher global disease activity (baseline SLEDAI-2K median 14 vs 12). Although these differences did not uniformly attain statistical significance due to the limited sample size, they underscore the necessity of focusing on longitudinal within-group trajectories rather than between-group efficacy comparisons in this analysis. Furthermore, the two groups differed in background immunosuppressant use: belimumab-treated patients received tacrolimus more frequently (80% vs 43.8%), whereas mycophenolate mofetil was more common in the telitacicept group (75% vs 40%). These baseline imbalances preclude any direct comparison of biological efficacy and reinforce that the belimumab arm serves solely as a descriptive reference for B-cell dynamics.

Table 1. Baseline characteristics of the patients.

Parameters	Telitacicept n=16	Belimumab n=5	P value
Demographic characteristics
Ages (years)	35.0 (29.5–45.0)	36.0 (27.5–41.0)	0.858
Female, n (%)	16 (100)	5 (100)	>0.999
BMI (kg/m2)	24.2 (21.2–28.1)	25.5 (19.5–28.9)	0.838
Disease-specific characteristics
Diseases duration of SLE (years)	7.5 (4.3–11.8)	11.0 (8.5–14.5)	0.093
SLEDAI-2K	14 (12–18)	12 (11–13)	0.060
Diseases duration of LN (years)	0.9 (0.3–4.3)	2.0 (0.8–4.5)	0.404
Laboratory indicators
24-hour urine protein (g/day)	2.17 (0.87–7.17)	4.79 (3.69–6.27)	0.275
ESR (mm/hour)	27 (18–52)	49 (29–61)	0.232
Low C3 and/or C4, n (%)	13 (81.25)	3 (60)	0.553
IgG (g/L)	9.4 (7.5–11.5)	7.7 (6.1–12.5)	0.660
ALB (g/L)	35.9 (29.7–37.7)	30.8 (25.9–35.0)	0.234
Scr (µmol/L)	57.0 (47.3–73.0)	47.0 (40.5–104.5)	0.587
eGFR (mL/min/1.73 m2) (n=13 vs 5)*	113.3 (84.3–140.4)	132.1 (70.1–172.0)	0.633
CD19+ B cells number (/µL) (n=15 vs 5)*	116 (56–224)	107 (81–271)	0.749
LN class, n (%)
Class Ⅲ/Ⅳ	4 (25)	1 (20)	>0.999
Class Ⅴ	2 (12.5)	3 (60)	0.063
Class Ⅲ/Ⅳ + Ⅴ	10 (62.5)	1 (20)	0.149
Extrarenal manifestations, n (%)
Mucocutaneous lesions	8 (50)	1 (20)	0.338
Arthritis	5 (31.25)	2 (40)	>0.999
Alopecia	2 (12.5)	0	>0.999
Vasculitis	1 (6.25)	0	>0.999
Pleuritic chest pain	2 (12.5)	0	>0.999
Haematological damage	1 (6.25)	2 (40)	0.128
Positive antibody, n (%)
ANA	15 (93.75)	5 (100)	>0.999
Anti-dsDNA (n=15 vs 5)*	12 (75)	2 (40)	0.131
Anti-dsDNA titre (IU/mL)	303.6 (220.6–395.9)	118.0 (44.4–316.2)	0.168
Anti-Sm (n=11 vs 5)*	3 (27.27)	4 (80)	>0.999
Anti-SSA (n=11 vs 5)*	8 (72.73)	4 (80)	0.106
Previous treatments, n (%)
Glucocorticoids	16 (100)	5 (100)	>0.999
Hydroxychloroquine	10 (62.5)	4 (80)	0.624
Mycophenolate mofetil	12 (75)	2 (40)	0.280
Tacrolimus	7 (43.75)	4 (80)	0.319
Daily glucocorticoid dose (mg)	15.0 (10.0–25.0)	20.0 (15.0–30.0)	0.197

Data are expressed as median (IQR, 25th–75th percentiles) for continuous variables, and as number (percentage) for categorical variables.

No missing data imputation was performed. Between-group comparisons were performed using the Mann-Whitney U test for continuous variables and Fisher’s exact test for categorical variables.

^*Indicates variables with missing baseline data, and the actual number of available cases used for statistical comparison is presented in parentheses.

ALB, albumin; anti-Sm, anti-Smith; anti-SSA, anti–Sjögren's-syndrome-related antigen A; BMI, body mass index; dsDNA, double-stranded DNA; eGFR, estimated glomerular filtration rate; ESR, erythrocyte sedimentation rate; LN, lupus nephritis; Scr, serum creatinine; SLEDAI-2K, SLE Disease Activity Index 2000.

Overall clinical efficacy of B-cell-targeted therapy

At week 24, 52.4% (11/21) and 38.1% (8/21) of the total cohort achieved the primary endpoint PERR and CRR, respectively (figure 2A, B). Assessment based on composite renal remission criteria (CR/PR/NR) also demonstrated a time-dependent increase in treatment response rates (figure 2C). The global disease activity, as measured by SLEDAI-2K score, decreased significantly from a median of 14.0 at baseline to 6.0 at week 24 (figure 2D).

Figure 2. Clinical efficacy of B-cell-targeted agents. (A–C) Changes in renal response outcomes, including PERR, CRR and the number of patients achieving CR/PR/NR. (D) Changes in SLEDAI-2K scores. (E) Dynamic changes in 24-hour urine protein during treatment (left) and distribution of proteinuria response at endpoint compared with baseline (right). (F–N) Changes in secondary clinical parameters over 24 weeks. (O) Changes in daily prednisone dosage. Data are presented as median with IQR based on available-case analyses without data imputation. Longitudinal within-group changes from baseline were evaluated using the Wilcoxon signed-rank test, with p values adjusted via the FDR method. *p<0.05; **p<0.01; ***p<0.001 versus baseline. Individual data points are slightly jittered horizontally to avoid overlapping and to ensure that the distribution of all patient scores is clearly visible. All data points accurately correspond to their scheduled visits (Weeks 0, 4, 12 and 24), and all baseline measurements represent the true pretreatment status. Alb, albumin; BLM, belimumab; CR, complete remission; CRR, complete renal response; dsDNA, double-stranded DNA; eGFR, estimated glomerular filtration rate; ESR, erythrocyte sedimentation rate; FDR, false discovery rate; NR, no response; PERR, primary efficacy renal response; PR, partial remission; Scr, serum creatinine; SLEDAI-2K, SLE Disease Activity Index 2000; TAI, telitacicept.

A marked reduction in 24-hour urine protein was observed, declining from a median of 3.79 g/day at baseline to 1.24 g/day at week 24, with 61.9% (13/21) of patients achieving at least a 50% reduction or normalisation by the endpoint (figure 2E). Significant improvements were also noted in inflammatory and immunological markers, including ESR, serum albumin, D-dimer, complement C3/C4, serum IgG levels and anti-dsDNA titres (figure 2F-K, N), while renal function (serum creatinine and eGFR) remained stable (figure 2L, M). Consequently, the median daily prednisone-equivalent dose was successfully tapered from 15.0 mg to 10.0 mg (figure 2O).

Exploratory observation of clinical response

While baseline differences (particularly the higher proportion of Class V LN and proteinuria in the BLM group) preclude direct efficacy comparisons, longitudinal trends were descriptively examined. No direct inferential statistics were performed between the two groups.

In the telitacicept group, a numerical trend toward clinical response was observed, with the PERR rate reaching 43.8% (7/16) as early as week 4 and increasing to 62.5% (10/16) by week 24. The belimumab group showed a trend as well, although in a smaller and more refractory cohort, with a PERR rate of 20% (1/5) at week 24 (online supplemental figure S2A). This descriptive trend remained consistent for CRR; telitacicept-treated patients achieved a rate of 50% (8/16), while a 0% (0/5) response was recorded in the belimumab reference arm (online supplemental figure S2B), likely reflecting its higher baseline proteinuric burden. Total renal remission (CR/PR) was achieved by 62.5% (10/16) and 40% (2/5) of patients in each group, respectively (online supplemental figure S2C).

Correspondingly, telitacicept treatment was associated with a marked and progressive reduction in 24-hour urine protein levels (online supplemental figure S2D). In the TAI group, median proteinuria decreased significantly from 2.17 g/day at baseline to 0.82 g/day by week 24 (p<0.01), with a noticeable decline observed as early as week 12. While the BLM reference group presented with a higher initial proteinuric burden, a numerical downward trend was also observed, with levels declining from a median of 4.79 g/day to 2.61 g/day at the end of the study period. Furthermore, both agents effectively reduced global disease activity, as evidenced by the evident decrease in SLEDAI-2K scores (online supplemental figure S2E).

The longitudinal trajectories of laboratory parameters further characterised the exploratory profiles within each agent table 2. Serum albumin levels exhibited a notable recovery over the 24-week period. Specifically, the telitacicept group reached a median albumin level of 41.6 g/L by the endpoint, reflecting a robust restoration of the glomerular filtration barrier, whereas the belimumab reference group showed improvement to a median of 36.9 g/L at week 24. Moreover, telitacicept was associated with a broader immunomodulatory profile, characterised by marked reductions in serum immunoglobulins (IgG, IgA and IgM) and a pronounced, early decline in anti-dsDNA titres starting from week 12, while these immunological parameters followed more gradual shifts or numerical fluctuations in the belimumab cohort. Collectively, these overall laboratory shifts highlight the profound immunomodulatory impact of dual BAFF/APRIL inhibition in this clinical setting.

Table 2. Clinical characteristics of patients with LN treated with telitacicept or belimumab over time.

Parameters and time points	Telitacicept (total n=16)	Belimumab (total n=5)	Nominal p value*
SLEDAI-2K
Baseline	14 (12–18)	12 (11–13)	0.060
Week 4	11 (8–16)**	10 (8–13)	0.361
Week 12	8 (6–10)***	8 (7–10)	0.885
Week 24	7 (6–9)***	6 (4–9)	0.413
24-hour urine protein (g/day)
Baseline	2.17 (0.87–7.17)	4.79 (3.69–6.27)	0.275
Week 4	2.16 (0.60–7.17)	4.75 (2.49–11.08)	0.109
Week 12	1.56 (0.50–3.67)**	3.69 (2.47–6.11)	0.091
Week 24	0.82 (0.50–3.38)**	2.61 (1.80–7.08)	0.091
C3 (g/L)
Baseline	0.80 (0.51–0.92)	0.82 (0.44–1.21)	0.732
Week 4 (n=14 vs 5)†	0.93 (0.80–1.03)**	1.01 (0.62–1.28)	0.978
Week 12 (n=14 vs 5)†	0.94 (0.80–1.00)‡	1.06 (0.75–1.24)	0.673
Week 24 (n=14 vs 5)†	1.02 (0.92–1.18)**	0.82 (0.76–1.12)	0.177
C4 (g/L)
Baseline	0.12 (0.05–0.20)	0.10 (0.03–0.25)	0.921
Week 4 (n=14 vs 5)†	0.15 (0.12–0.23)‡	0.21 (0.09–0.27)	0.983
Week 12 (n=14 vs 5)†	0.15 (0.13–0.28)**	0.22 (0.14–0.28)	0.512
Week 24 (n=14 vs 5)†	0.17 (0.15–0.26)**	0.20 (0.12–0.22)	0.485
ESR (mm/hour)
Baseline	27 (18–52)	49 (29–61)	0.232
Week 4 (n=12 vs 4)†	21 (20–43)‡	38 (37–38)	0.152
Week 12 (n=12 vs 4)†	28 (22–30)	18 (18–36)	0.428
Week 24 (n=12 vs 4)†	13 (10–22)***	19 (15–22)	0.330
Scr (µmol/L)
Baseline	57.0 (47.3–73.0)	47.0 (40.5–104.5)	0.587
Week 4 (n=15 vs 5)†	58.0 (49.0–88.0)	60.0 (43.0–101.5)	>0.999
Week 12 (n=15 vs 5)†	58.0 (47.0–79.0)	57.0 (39.5–110.5)	0.717
Week 24	61.0 (48.0–81.8)	57.0 (39.5–109.5)	0.589
eGFR (mL/min/1.73 m2)
Baseline (n=13 vs 5)†	113.3 (84.3–140.4)	132.1 (70.1–172.0)	0.633
Week 4 (n=12 vs 5)†	112.6 (64.5–129.8)	99.7 (69.4–159.6)	0.940
Week 12 (n=12 vs 5)†	109 (73.0–150.5)	105.8 (64.0–174.8)	0.740
Week 24 (n=13 vs 5)†	103.5 (68.8–146.6)	105.8 (65.6–174.8)	0.633
ALB (g/L)
Baseline	35.9 (29.7–37.7)	30.8 (25.9–35.0)	0.234
Week 4	36.7 (32.0–41.2)***	34.3 (26.7–37.4)	0.174
Week 12	38.3 (35.1–40.9)***	37.1 (28.4–37.9)	0.129
Week 24	41.6 (38.5–43.5)***	36.9 (31.6–38.5)	0.032
D-Dimer (mg/L)
Baseline (n=13 vs 5)†	0.57 (0.23–1.66)	0.75 (0.32–3.44)	0.633
Week 4 (n=10 vs 5)†	0.52 (0.17–0.85)‡	0.54 (0.39–0.94)	0.572
Week 12 (n=12 vs 5)†	0.24 (0.17–0.55)**	0.56 (0.11–0.62)	0.898
Week 24 (n=13 vs 5)†	0.26 (0.12–0.49)**	0.30 (0.20–0.49)	0.924
IgG (g/L)
Baseline	9.4 (7.5–11.5)	7.7 (6.1–12.5)	0.660
Week 4 (n=15 vs 5)†	8.5 (7.2–9.3)***	7.8 (5.2–9.0)	0.314
Week 12 (n=15 vs 5)†	7.8 (6.5–8.8)**	8.3 (5.6–8.7)	0.850
Week 24	7.3 (6.6–9.0)**	7.1 (5.8–8.5)	0.707
IgM (g/L)
Baseline	0.64 (0.34–1.08)	0.35 (0.34–0.96)	0.704
Week 4 (n=15 vs 5)†	0.39 (0.28–0.84)**	0.39 (0.33–0.87)	0.627
Week 12 (n=15 vs 5)†	0.37 (0.25–0.80)**	0.37 (0.29–0.79)	0.719
Week 24	0.29 (0.20–0.69)***	0.30 (0.26–0.66)	0.733
IgA (g/L)
Baseline	2.00 (1.4–2.94)	2.60 (1.27–3.55)	0.768
Week 4 (n=15 vs 5)†	1.46 (1.30–2.00)***	2.38 (1.27–2.78)	0.497
Week 12 (n=15 vs 5)†	1.26 (1.01–1.26)***	2.41 (1.23–2.75)	0.168
Week 24	1.13 (0.95–2.09)***	2.34 (1.19–2.78)	0.073
WBC (10ˆ9 /L)
Baseline	7.3 (6.0–9.7)	6.8 (4.3–10.0)	0.602
Week 4	8.2 (6.1–9.5)	6.4 (5.2–9.7)	0.707
Week 12	6.8 (5.2–8.5)	6.2 (5.4–8.4)	0.983
Week 24	6.4 (4.5–9.7)	7.3 (5.2–9.9)	0.719
PLT (10ˆ9 /L)
Baseline	240 (192–291)	142 (106–279)	0.174
Week 4	263 (200–346)	146 (136–276)	0.039
Week 12	265 (208–287)	204 (148–295)	0.124
Week 24	242 (193–201)	166 (152–321)	0.431
Hb (g/L)
Baseline	106 (97–120)	91 (66–118)	0.248
Week 4	114 (84–128)‡	114 (84–128)	0.702
Week 12	115 (105–125)‡	122 (86–135)	0.857
Week 24	120 (108–132)**	116 (89–126)	0.320
ANA titre
Baseline	1:1000 (1:320–1:1000)	1:1000 (1:320–1:3200)	0.458
Week 4 (n=2 vs 0)†	1:1000 (—, —)	NA	—
Week 12 (n=12 vs 4)†	1:320 (1:160–1:1000)	1:1000 (1:320–1:3200)	0.397
Week 24 (n=15 vs 4)†	1:320 (1:100–1:320)‡	1:1000 (1:320–1:3200)	0.078
Anti-dsDNA (IU/mL)
Baseline (n=15 vs 5)†	303.6 (220.6–395.9)	118.0 (44.4–316.2)	0.168
Week 4 (n=4 vs 2)†	121.9 (26.8–225.5)	293.9 (119.1–468.7)	0.533
Week 12 (n=12 vs 4)†	131.2 (59.5–282.9)‡	121.8 (71.2–371.3)	0.862
Week 24 (n=14 vs 4)†	138.6 (77.9–252.1)**	99.7 (48.9–359.9)	>0.999
Glucocorticoid dose (mg/day)
Baseline	15.0 (10.0–25.0)	20.0 (15.0–30.0)	0.197
Week 4	15.0 (10.0–25.0)‡	15.0 (11.3–22.5)	0.556
Week 12	11.3 (8.1–15.0)**	10.0 (10.0–20.0)	0.673
Week 24	8.0 (6.6–10.0)***	10.0 (7.5–18.8)	0.274

Data are presented as median (IQR) based on available cases without missing data imputation.

^*P values shown in this column are nominal and exploratory only, derived from unadjusted Mann-Whitney U tests. They are not corrected for multiple comparisons and should not be interpreted as confirmatory evidence of between-group differences, given the non-randomised design and small sample size. No formal inferential conclusions are drawn from these values.

^†Indicates variables with missing data during follow-up visits, with the actual number of available cases shown in parentheses.

^‡*Indicates significant within-group changes compared to baseline in the telitacicept group (Wilcoxon signed-rank test with FDR-adjusted p values: *p<0.05; **p<0.01; ***p<0.001); the belimumab arm is shown for descriptive reference only.

ALB, albumin; dsDNA, double-stranded DNA; eGFR, estimated glomerular filtration rate; ESR, erythrocyte sedimentation rate; FDR, false discovery rate; Hb, haemoglobin; LN, lupus nephritis; NA, not assessed; PLT, platelet; Scr, serum creatinine; SLEDAI-2K, SLE Disease Activity Index 2000; WBC, white blood cell.

Both agents were generally well tolerated within the 24-week observation period. Transient infections, including upper respiratory and urinary tract infections, were reported in both groups but required no permanent treatment discontinuation. No unexpected safety signals or serious adverse events were identified. However, given the small sample size and short follow-up, definitive conclusions on long-term safety cannot be drawn.

Modulation of B-cell lineage and homeostatic subsets by B-cell-targeted therapies

Analysis of lymphocyte frequencies confirmed that both agents selectively targeted the B-cell lineage. Due to the exploratory and real-world design of this study, the number of patients analysed at each time point (W0, W4, W12, W24) was n=16, 9, 14, 16 for the telitacicept group and n=5, 4, 4, 5 for the belimumab reference group, respectively. While CD3⁺ T cell proportions exhibited an initial upward trend before reaching a steady state in both treatment groups (figure 3A), total CD19⁺ B-cell frequencies and absolute counts followed a progressive downward trajectory (figure 3B, online supplemental figure S3A). At the subset level, both therapies effectively contracted the BAFF-dependent naïve B-cell compartment in terms of both frequency and absolute number (figure 3C, online supplemental figure S3B).

Figure 3. Distinct modulation of B-cell subsets by dual BAFF/APRIL versus single BAFF blockade. (A–I) Temporal changes in the compositional frequencies of circulating lymphocyte and B-cell subsets in patients treated with TAI (red) and BLM (blue). Frequencies are expressed as the percentage of live lymphocytes (for CD3+ T cells and CD19+ B-cells) or as the percentage of the total CD19+ B-cell pool (for downstream B-cell and antibody secreting cell subsets). Rare populations experiencing trans-order-of-magnitude depletion are plotted on a Log10 Y-axis, with zero values assigned a nominal LOD baseline strictly for logarithmic visualisation. Data are presented as median (IQR) based on available-case analyses. Longitudinal within-group changes from baseline were evaluated using the Wilcoxon signed-rank test, with p values adjusted via the FDR method. *p<0.05; **p<0.01; ***p<0.001 versus baseline. Individual data points are slightly jittered horizontally to avoid overlapping and to ensure that the distribution of all patient scores is clearly visible. All data points accurately correspond to their scheduled visits (Weeks 0, 4, 12 and 24), and all baseline measurements represent the true pretreatment status. No statistical comparisons between treatment groups are presented due to the small and non-randomised study design; the BLM arm is shown for descriptive reference only. BLM, belimumab; FDR, false discovery rate; LOD, limit of detection; TAI, telitacicept.

Among conventional non-pathogenic subsets, the frequencies and absolute counts of unswitched memory (USWM) B-cell remained largely stable or showed minimal fluctuation in both groups (figure 3D, online supplemental figure S3C). Notably, DN1 cells exhibited a transient elevation in the telitacicept group, with both frequency and absolute counts peaking at week 12 before declining toward the endpoint (figure 3F, online supplemental figure S3E). This pattern likely reflects a temporary relative enrichment or a compensatory response as other sensitive B-cell compartments were preferentially cleared. Similarly, while absolute counts of switched memory (SWM) B-cells remained stable (online supplemental figure S3D), their relative frequency significantly increased in the telitacicept group (figure 3E). This phenomenon represents a passive proportional reshaping of the residual B-cell pool following the profound contraction of other populations, further distinguishing this remodelling process from an active immune reset.

Differential targeting of the pathogenic extrafollicular axis by telitacicept and belimumab

The modulation of the pathogenic extrafollicular axis, however, revealed distinct pharmacodynamic patterns between the two agents. Although both agents induced an initial decline in the proportion of DN2 (CD11c⁺CXCR5⁻CD21⁻) B-cells, telitacicept achieved a more sustained suppression through week 24, whereas the belimumab reference group showed a numerical recovery toward baseline levels at the study endpoint (figure 3G). The most striking divergence occurred in the effector compartments; telitacicept induced a profound depletion of CD27⁺ plasmablasts and CD138⁺ mature plasma cells, whereas these subsets persisted or exhibited a relative expansion in the belimumab cohort (figure 3H, I). Specifically, the frequency of plasmablasts in the telitacicept group trended lower than that in the belimumab group by week 24 (figure 3H), suggesting a potential advantage of dual blockade in eliminating active effector populations; however, direct statistical comparison was not performed due to the small and imbalanced sample sizes.

Absolute cell counts further corroborated these frequency-based observations, highlighting differing kinetics within the extrafollicular pathway (online supplemental figure S3). Both agents successfully induced a reduction in DN2 B-cell absolute counts (online supplemental figure S3F), yet with differing dynamics. Telitacicept induced a rapid contraction as early as week 4, followed by sustained suppression through week 24, achieving high statistical significance versus baseline. Conversely, the belimumab reference group exhibited a gradual, steady decline over the 24-week period. This divergence likely reflects that dual BAFF/APRIL blockade by telitacicept rapidly interrupts survival signals for activated DN2 cells, driving early depletion, whereas single BAFF inhibition by belimumab may progressively limit DN2 recruitment or maturation over time. A more pronounced pharmacological divergence was observed within the effector compartments. Telitacicept drove a profound and progressive decline in CD27⁺ plasmablasts and CD138⁺ mature plasma cells, reaching significantly lower absolute counts by week 24 (online supplemental figure S3G–H). Conversely, these APRIL-dependent populations remained largely refractory to belimumab. Consequently, absolute plasmablast counts in the telitacicept group showed a descriptive tendency toward lower values compared with the belimumab group at the study endpoint (online supplemental figure S3G); although formal statistical inference was not applied given the exploratory nature of the between-group comparison. This difference potentially highlights the dependence of plasmablasts on APRIL, which can be effectively neutralised by the dual inhibition of telitacicept.

B-cell dynamics as correlates and early predictors of telitacicept treatment response

To assess the clinical relevance of B-cell modulation by telitacicept, we performed Spearman correlation analyses between B-cell subsets and disease activity markers (figure 4A). The naïve B-cell compartment correlated positively with SLEDAI-2K scores and 24-hour proteinuria in both frequency and absolute count, likely reflecting an expanded precursor pool available for recruitment into pathogenic pathways during disease flares. The downstream pathogenic extrafollicular axis, including DN2 cells, plasmablasts and mature plasma cells, also correlated positively with disease activity, underscoring its role in renal impairment. In contrast, SWM frequency exhibited a negative correlation with disease activity, serving as a hallmark of B-cell landscape remodelling. However, absolute counts of both SWM and USWM B-cells correlated positively with SLEDAI-2K scores and 24-hour proteinuria, reflecting global B-cell expansion during flares which is subsequently reversed by telitacicept-mediated comprehensive depletion.

Figure 4. Targeted ablation of pathogenic B-cell subsets by telitacicept drives deep clinical remission in lupus nephritis. (A) Spearman correlation heatmaps evaluating the associations between B-cell subset profiles (compositional frequencies and absolute counts) and clinical parameters (SLEDAI-2K and 24-hour urine protein). Colour gradients represent Spearman's rank correlation coefficients (ρ). (B) Comparative changes in the compositional frequencies of specific B-cell subsets in telitacicept responders (PERR, n=10; red) versus non-responders (N-PERR, n=6; green). (C) Alterations in the absolute counts of corresponding B-cell subsets in the PERR and N-PERR groups. (D) ROC curves evaluating the associative capacity of week-12 dynamic reductions (percentage change from baseline) in specific B-cell subsets for achieving PERR at week 24. The AUC, 95% CIs and p values are displayed. Data in panels B and C are presented as median (IQR) based on available-case analyses. Longitudinal within-group changes from baseline were evaluated using the Wilcoxon signed-rank test, with p values adjusted via the FDR method. *p<0.05; **p<0.01; ***p<0.001 versus baseline. Individual data points are slightly jittered horizontally for optimal visualisation. Between-group comparisons were not performed for the data shown in panels B and C given the exploratory nature of the analysis. Notably, the AUC values reported in panel D are derived from a small cohort (n=16) and should be interpreted as exploratory correlates rather than definitive clinical predictors. AUC, area under the curve; DN, double-negative; FDR, false discovery rate; PERR, primary efficacy renal response; ROC, receiver operating characteristic; SLEDAI-2K, SLE Disease Activity Index 2000; SWM, switched memory; USWM, unswitched memory.

These clinical correlations were consistent with longitudinal observations stratified by week-24 PERR status (responders, n=10; non-responders, n=6; figure 4B, C and online supplemental figure S4). Both groups exhibited substantial reductions in naïve B-cell counts, reflecting a consistent pharmacological effect of telitacicept across the cohort. However, clinical remission was distinguished by deeper eradication of the downstream pathogenic axis. Responders achieved deep elimination of plasmablasts and mature plasma cells in both frequency and absolute count by week 24, coupled with sustained reductions in DN2 absolute counts. This comprehensive ablation was accompanied by relative enrichment of SWM frequency. In contrast, non-responders failed to suppress these primary effector lineages. Accessory subsets such as USWM and DN1 cells showed comparable trajectories regardless of outcome.

Finally, ROC analysis demonstrated that early immunological dynamics (percentage reduction from baseline) at week 12 could be associated with week-24 clinical response (figure 4D). Week-12 reductions in plasmablasts and mature plasma cells exhibited high discriminatory power, with AUC of 0.836 (p=0.036) and 0.964 (p=0.004), respectively. Meanwhile, the early contraction of naïve B-cells also demonstrated high relevance to clinical outcomes (AUC=0.909, p=0.004). Given the small sample size (n=16 in the telitacicept group), these AUC estimates are exploratory and likely overfitted; they require external validation in larger independent cohorts before any clinical application. With that caveat, these findings suggest that the early neutralisation of both precursors and the downstream extrafollicular axis may serve as early immunological correlates of sustained renal response to telitacicept.

Discussion

In this prospective cohort study, we characterised the progressive B-cell landscape remodelling and associated clinical renal responses in patients with LN receiving dual BAFF/APRIL inhibitor telitacicept, using a belimumab-treated cohort as a descriptive pharmacological reference. This enhanced clinical efficacy is paralleled by a unique immunomodulatory profile, specifically the profound depletion of the extrafollicular axis and the subsequent remodelling of B-cell homeostasis. Furthermore, we established that the early reductions in plasmablasts, mature plasma cells and naïve B-cells may serve as early correlates associated with treatment response to telitacicept.

Our findings on the clinical efficacy of belimumab are consistent with prior large-scale trials and real-world studies, which report PERR rates of approximately 40% at 6 months and often observe delayed responses in refractory patients.^{29 34} Notably, the belimumab group in our study presented with a higher proportion of Class V LN and proteinuria at baseline, which likely influenced the observed clinical trajectories. Despite these baseline differences, the pattern of B-cell subset modulation we observed—sustained reduction of naïve B-cells and limited impact on mature effector cells—aligns well with the known mechanisms of BAFF inhibition.^{14 35 36} This consistency validates our experimental approach and provides a reliable pharmacological baseline for characterising the distinct effects of the novel agent telitacicept.

The pronounced renal response rates and rapid reduction in proteinuria and anti-dsDNA antibodies observed with telitacicept underscore the therapeutic advantage of dual BAFF/APRIL blockade. The compelling mechanistic insight from our study is the sustained depletion of plasmablasts and mature plasma cells exclusively by telitacicept. Notably, we detected a substantial burden of circulating CD138⁺ mature plasma cells at baseline—a pathological phenomenon likely reflecting bone marrow niche saturation and spillover during severe flares.^{17 37 38} Such effective clearance reflects the distinct receptor–ligand dynamics of the BAFF/APRIL system and underscores the depth of B-cell compartmental redistribution unique to dual blockade. Plasmablasts and plasma cells highly express BCMA and TACI, receptors for which APRIL has a markedly higher affinity than BAFF.³⁹ By concurrently neutralising APRIL, telitacicept directly disrupts the critical survival pathway for these ASCs, a mechanism that selective BAFF inhibition with belimumab cannot fully replicate.^{18 40 41} This direct targeting of plasmablasts and mature plasma cells likely underlies the more rapid serological and clinical improvements observed with telitacicept.

A critical translational contribution of our work is the identification of dynamic B-cell biomarkers associated with telitacicept response through both absolute counts and relative frequencies. We observed a robust negative correlation between SWM frequency and disease severity. This is likely because telitacicept preferentially depletes BAFF/APRIL-dependent naive B cells and extrafollicular pathogenic cells, leading to a passive relative enrichment of the existing SWM pool. Correspondingly, a reduction in the B-cell pool accompanied by an increase in SWM frequency may serve as a marker of deep clinical remission. However, this phenomenon is fundamentally distinct from the ‘immune reset’ observed in cell-based interventions. Typically, CD19-targeted chimeric antigen receptor-T therapy induces a profound, systemic depletion of all CD19⁺ B cells—including both naïve and memory subsets—leading to a subsequent immune reconstitution characterised by a predominantly naïve B-cell repertoire.⁴² In contrast, this pattern observed with telitacicept treatment represents a passive proportional reshaping of the residual pool rather than an active ‘immune reset’, as evidenced by the lack of complete SWM B-cell ablation. This distinction further justifies the use of ‘B-cell landscape remodeling’ to describe the pharmacodynamic impact of dual BAFF/APRIL blockade.

Meanwhile, our work identifies early immunological shifts at week-12 as reliable early on-treatment biomarkers associated with long-term clinical outcomes. We demonstrated that the magnitude of reduction in mature plasma cells (AUC=0.964), plasmablasts (AUC=0.836) and naïve B-cells (AUC=0.909) provides a ‘molecular window’ into the likelihood of achieving a renal response by week 24. These AUC values underscore that the early B-cell depletion is not merely a transient pharmacodynamic shift, but rather a significant remodelling of the B-cell landscape associated with deep remission. The profound depletion of plasmablasts and mature plasma cells—the primary source of autoantibodies—directly reflects the robust systemic clearance of the pathogenic effector cells driving renal damage. Crucially, the strong association observed between naïve B-cell and clinical response contraction suggests that limiting the precursor supply to the extrafollicular DN2-plasmablast axis is as vital as the direct neutralisation of the effector cells themselves. Thus, early, coordinated suppression of both the upstream precursor pool and the downstream effector axis may represent a key biological feature of sustained renal improvement. Despite the exploratory nature of our study, these findings pave the way for a more personalised treatment approach, where on-treatment B-cell monitoring could assist in therapeutic decisions, potentially identifying non-responders early for intervention.

Several limitations of this study must be acknowledged. First, the non-randomised, real-world design resulted in imbalanced group sizes (16 vs 5) and notable differences in baseline disease characteristics (eg, proteinuria, LN class) as well as background immunosuppressant use (tacrolimus predominance in the belimumab group). These confounders preclude definitive efficacy comparisons; the belimumab group therefore served only as a descriptive pharmacological reference. Importantly, while conventional immunosuppressants often result in gradual and imprecise B-cell impairments, the profound and rapid remodelling of the extrafollicular axis observed here appears to be a specific pharmacodynamic effect of telitacicept. This suggests that the observed immunological homeostatic repopulation was primarily driven by dual BAFF/APRIL inhibition rather than generalised B-cell shifts associated with standard-of-care therapy. Second, to avoid artificially exaggerating significance in a small cohort, we rigorously avoided missing data imputation. Third, differing administration routes (subcutaneous vs intravenous) may influence pharmacokinetics, though assessment was beyond this study’s scope. Furthermore, while the presence of anti–Sjögren's-syndrome-related antigen A (anti-SSA) antibodies in many patients suggests a high-interferon (IFN) background, we did not directly measure IFN gene signature.⁴³ However, the observed depletion of IFN-responsive DN2 B-cells and the reduction in autoantibody titres provide indirect evidence for telitacicept’s potential impact on the IFN axis and warrants further investigations. Finally, the 24-week peripheral blood monitoring window cannot capture depletion of deep-tissue-resident long-lived plasma cells; our conclusions are thus confined to the circulating extrafollicular axis. Validation in larger, randomised cohorts is warranted.

In conclusion, through a prospective, mechanistic observation, we demonstrate that telitacicept, through dual BAFF/APRIL inhibition, exerts a distinct immunological profile characterised by profound extrafollicular ablation. We further establish a link between specific B-cell dynamics and treatment efficacy, proposing specific circulating B-cell subsets as promising early indicators associated with clinical response. These insights not only elucidate the mechanistic basis of dual BAFF/APRIL inhibition but also provide a rational framework for optimising treatment strategies and advancing towards personalised management of LN.

Data availability statement

Data are available upon reasonable request.