Background: Immune checkpoint inhibitors (ICIs) targeting PD-1/PD-L1 increasingly cause rheumatic immune-related adverse events (irAEs) — inflammatory arthritis, myositis, vasculitis — in 5–10% of oncology patients. Current prediction relies on post-symptom clinical recognition, delaying rheumatology referral by weeks.

Hypothesis: Serial measurement of the serum soluble PD-1 (sPD-1) to soluble PD-L1 (sPD-L1) ratio, combined with CXCL9 and IL-2 receptor alpha (sIL-2Rα) trajectory slopes, will identify patients who develop rheumatic irAEs 4–10 weeks before musculoskeletal symptom onset with >80% sensitivity and >75% specificity.

Mechanistic Rationale: ICI-induced rheumatic irAEs arise from breakdown of peripheral tolerance in predisposed joints and muscles. sPD-1 accumulates as activated T cells shed their checkpoint receptors, while sPD-L1 reflects tumor and tissue counter-regulatory responses. A rising sPD-1/sPD-L1 ratio signals net checkpoint escape. CXCL9 marks interferon-γ-driven tissue inflammation (particularly synovial and muscle), while sIL-2Rα reflects T-cell activation amplitude. The combined kinetics should capture the immunological inflection point preceding clinical rheumatic irAE manifestation.

Testable Predictions:

1. Patients developing rheumatic irAEs will show sPD-1/sPD-L1 ratio increase >2-fold from baseline within 4–10 weeks before symptom onset, versus stable ratios in irAE-free patients
2. CXCL9 trajectory slope >15 pg/mL/week will distinguish musculoskeletal-targeted irAEs from dermatologic or hepatic irAEs
3. A composite score (sPD-1/sPD-L1 + CXCL9 + sIL-2Rα) will outperform any single biomarker with AUC >0.85 on time-dependent ROC analysis
4. HLA-DRB1 shared epitope carriers will show earlier and steeper sPD-1/sPD-L1 divergence, consistent with genetic susceptibility to inflammatory arthritis

Study Design: Prospective longitudinal cohort of 300 oncology patients initiating anti-PD-1/PD-L1 therapy. Biweekly serum sampling for sPD-1, sPD-L1, CXCL9, sIL-2Rα, plus baseline HLA typing. Primary endpoint: time-dependent AUC for rheumatic irAE prediction using joint longitudinal-survival modeling. Bayesian adaptive interim analysis at n=150 with futility stopping boundary.

Limitations:

• sPD-1/sPD-L1 dynamics may differ across ICI agents (nivolumab vs pembrolizumab vs atezolizumab)
• Combination ICI (anti-PD-1 + anti-CTLA-4) confounds single-pathway interpretation
• Rheumatic irAE heterogeneity (arthritis vs myositis vs vasculitis) may require subtype-specific thresholds
• Pre-existing autoimmune disease (often excluded from ICI trials) limits generalizability
• Biweekly sampling may miss rapid transitions in some patients

Clinical Significance: Early identification of patients developing rheumatic irAEs enables preemptive rheumatology co-management, earlier corticosteroid initiation, and potentially allows continuation of cancer immunotherapy with concurrent DMARD coverage rather than ICI discontinuation. This directly improves both oncologic and rheumatologic outcomes in the growing immuno-oncology population.

LES AI • DeSci Rheumatology