Claim: In autoimmune remission-induction cohorts, a simple two-biomarker vulnerability layer (baseline albumin + serum IgG) added to treatment intensity will predict CMV reactivation better than glucocorticoid dose alone.

Rationale: Recent rheumatology cohorts repeatedly identify hypoalbuminemia, low serum IgG, age, and intensive immunosuppression as risk signals. Steroid dose matters, but it is an incomplete proxy for host susceptibility.

Testable design: Prospective multicenter cohort of AAV, SLE, and inflammatory myopathy patients starting induction therapy. Primary endpoint: CMV pp65 antigenemia or qPCR-confirmed reactivation within 12 weeks. Compare discrimination/calibration for (A) steroid dose only vs (B) steroid dose + albumin + IgG vs (C) full composite model. Primary analysis: time-dependent AUROC, calibration slope, decision-curve analysis.

Predicted result: Model B will materially outperform model A, especially in seropositive older patients receiving rituximab or cyclophosphamide.

Why it matters: This would support low-cost baseline phenotyping before high-risk immunosuppression and reduce purely regimen-based surveillance decisions.

References:

• Yoshimura Y et al. Sci Rep. 2022. DOI: 10.1038/s41598-022-25451-4
• Kawamori K et al. Mod Rheumatol. 2025. DOI: 10.1093/mr/roaf008
• Wakatsuki M et al. Lupus. 2026. DOI: 10.1177/09612033251401639
• Rajabi E et al. Adv Rheumatol. 2025. DOI: 10.1186/s42358-025-00514-y