Short-term antiproteinuric effects of disease-modifying therapies for IgA nephropathy: a systematic review and Bayesian mechanism-class indirect comparison

Technical Summary: Short-term Antiproteinuric Effects of Disease-Modifying Therapies for IgA Nephropathy

Problem Statement
Primary IgA nephropathy (IgAN) remains the most common primary glomerulonephritis globally, often progressing to kidney failure despite optimized renin–angiotensin system (RAS) inhibition. While early proteinuria reduction predicts long-term kidney function decline and has facilitated accelerated drug approvals, the therapeutic landscape has expanded rapidly with multiple mechanistically distinct agents. These include targeted-release corticosteroids, endothelin-pathway inhibitors, complement inhibitors, and APRIL/BAFF-axis therapies. A critical gap exists: no randomized trial has directly compared two active disease-modifying agents against each other. Consequently, clinicians must select between therapies with unmeasured relative effects and varying maturities of long-term kidney outcome data. This review addresses the need to estimate short-term antiproteinuric effects by mechanism class while explicitly defining the limits of interpretability for comparative claims in the absence of head-to-head evidence.

Methodology
This systematic review and Bayesian mechanism-class indirect comparison adhered to PRISMA 2020 guidelines. The authors searched MEDLINE/PubMed, Europe PMC, ClinicalTrials.gov, and other registries through June 24, 2026.

• Eligibility: Parallel-group randomized trials in adults with biopsy-confirmed primary IgAN were included if they reported a treatment-to-control urine protein-to-creatinine ratio (UPCR) geometric mean ratio (GMR) at 26–39 weeks.
• Data Harmonization: Effects were analyzed on the log-GMR scale. The authors carefully preserved the specific confidence levels reported by individual trials (e.g., 96.5% for VISIONARY, 90% for SANCTUARY) rather than forcing them into a standard 95% interval, to avoid misstating precision.
• Statistical Model: A Bayesian random-effects model was employed with mechanism class as the node. The model assumed exchangeability for short-term UPCR response within a shared pathway but explicitly did not assume therapeutic interchangeability of agents within a class. Priors were set as independent Normal(0, 1²) for class effects and half-Normal(0, 0.10²) for heterogeneity. The posterior was derived via exact numerical integration (quadrature) rather than Markov chain Monte Carlo.
• Bias and Confidence: Risk of bias was assessed using RoB 2, and confidence in the evidence was evaluated using CINeMA. Crucially, confidence was rated separately for "class-versus-control" estimates versus "active-versus-active" contrasts.
• Network Structure: The network was connected solely through a shared control anchor (optimized RAS inhibition, placebo, or active RAS inhibitor). No closed loops existed, rendering node-splitting and design-by-treatment inconsistency models structurally inestimable.

Key Contributions

• Mechanism-Class Estimation: The study synthesized 10 randomized contrasts involving 2,036 participants across four mechanism classes and nine agent nodes to provide class-level estimates of short-term antiproteinuric efficacy.
• Explicit Interpretive Boundaries: The paper distinguishes between class-level indirect estimates and head-to-head superiority. It clarifies that while ranking is possible, active-active comparisons are entirely indirect and subject to transitivity assumptions that cannot be empirically tested without closed loops.
• Differentiation of Surrogate vs. Outcome Maturity: The analysis highlights the disconnect between the strongest short-term surrogate signals (proteinuria reduction) and the maturity of long-term kidney preservation data, noting that agents with the highest proteinuria reduction (APRIL/BAFF-axis) currently lack the mature kidney outcome data available for agents like nefecon, sparsentan, and atrasentan.
• Methodological Rigor in Indirect Comparison: By using a joint Bayesian model rather than separate pairwise meta-analyses, the study allows sparse classes to borrow strength through a common heterogeneity parameter while propagating uncertainty rather than discarding it.

Results
The analysis identified four mechanism classes with the following modeled short-term UPCR reductions (Geometric Mean Ratio [GMR] and 95% Credible Interval [CrI]):

1. APRIL/BAFF-axis therapy: Largest reduction (GMR 0.50, 95% CrI 0.44–0.56; 50.5% reduction). Probability of being the best agent was 97.9%.
2. Endothelin-pathway therapy: GMR 0.62 (0.54–0.71; 38.3% reduction).
3. Complement inhibition: GMR 0.65 (0.55–0.77; 35.3% reduction).
4. Targeted-release corticosteroid: GMR 0.73 (0.61–0.87; 27.5% reduction).

• Risk of Bias: Seven trials were rated as low risk of bias; three had some concerns.
• Confidence (CINeMA): Confidence in class-versus-control estimates was rated as moderate. However, confidence in active-versus-active contrasts was rated as low to very low due to the entirely indirect nature of the comparisons and the inability to test for incoherence.
• Sensitivity Analyses: The APRIL/BAFF-axis class remained top-ranked across all prespecified sensitivity analyses (including variations in heterogeneity priors, exclusion of sparsentan, and restriction to low-risk trials).

Significance and Claims
The paper claims that APRIL/BAFF-axis therapy produced the largest short-term antiproteinuric signal available in the randomized evidence up to June 2026. However, the authors strictly qualify this finding:

• Not Evidence of Superiority: The result is a class-level indirect estimate, not evidence of head-to-head superiority or within-class interchangeability.
• Clinical Implication: Treatment selection should not rely solely on this surrogate ranking. Clinicians must weigh the "asymmetry" between the strong short-term proteinuria signal of APRIL/BAFF-axis agents and the more mature kidney-outcome data for other agents (nefecon, sparsentan, atrasentan, iptacopan).
• Future Needs: The evidence map identifies a critical need for active-active or platform trials, longer follow-up, harmonized kidney endpoints, and standardized reporting of safety and treatment burden to close the gap between surrogate response and kidney preservation.

The review concludes that while the comparative rankings generate hypotheses, they should not substitute for head-to-head trials or individualized treatment selection based on the full spectrum of efficacy, safety, and monitoring burden.